Ravi Mhatre

1. Snapshot and testable edge thesis

Edge thesis. Mhatre's public record is most consistent with an ability to recognize a technical architecture change before its enterprise value is obvious, identify a narrow urgent wedge, back a founder/team capable of adapting across product, GTM and platform phases, and re-underwrite the relationship through staged proof. The direct component rules are often strong; their combined effectiveness is only [researcher inference · moderate] because original memos, decision gates, function logs, the negative denominator and cash flows are missing.

The edge is testable rather than honorific. Test recognition with a dated pre-consensus memo naming architecture and buyer constraint; selection with a complete opportunity/pass/invest cohort matched by stage/vintage; support with contemporaneous board/recruit/follow-on actions tied to milestones; and economics with security-level vehicle cash flows after removing the largest one and two outcomes. Failure of any link weakens the combined mechanism. Profile test [researcher inference · strong]

Evidence ceiling. Forty selected writings, 31 appearance families and 91 live profile cards create a deep public teaching record, not a full opportunity/pass/write-off or return denominator. Audited Mhatre-only alpha is not established. [researcher inference · strong]

2. Core philosophy

• Technology is a language, not the investment conclusion. Translate a capability into the workflow constraint it removes, the buyer who cares and the feedback that proves value. Stanford, 00:00–08:31 [investor-stated · strong]
• Enter through an exact wedge. Grand vision matters only when a narrow first product can earn adoption and learning. Nutanix, MuleSoft and AppDynamics illustrate the rule retrospectively. Nutanix essay [researcher inference · strong]
• Underwrite adaptation, not a static founder type. The skills for product proof, GTM repeatability and platform expansion differ; the board/recruiting job changes with the phase. AppDynamics talk note [retrospective witness · strong]
• Use feedback loops instead of static prediction. Technical change makes long-term forecasts fragile; customer/product data should update conviction and capital. Newcomer, 07:45–11:10 [investor-stated · strong]
• Regime matters. The same tactic behaves differently under dot-com, GFC, cheap-capital, reset and AI “gravity”; preserve the rule while changing milestones, financing and pace. Talks evolution [researcher inference · strong]
• The firm is a capability, not personal credit. Multi-partner teams, advisers, vehicles and coauthors enable sourcing/diligence/support; record them rather than collapsing them into a famous partner. Decision-rights matrix [documented behavior · strong]

3. Sourcing

The visible sourcing system combines technical networks, Stanford/Bessemer/WPG/Lightspeed relationships, repeat founders, firm partners, writing/events and global institutional reach. Biographical/network boundary · Formation and people map · 2006 India-office route [researcher inference · strong] The 91 cards do not expose a funnel or conversion rate, and card association is not sourcing proof.

A usable source ledger should record:

1. cold / founder / operator / partner / portfolio / university / bank / thematic source;
2. first contact date and technical inflection under discussion;
3. which adviser/vehicle had mandate;
4. source, sponsor, diligence owner, IC authority and board candidate separately; and
5. all declined/parked opportunities, not only investments. [researcher inference · strong]

Repeat-founder access can shorten trust formation and improve reference quality; ThoughtSpot also shows its danger—familiarity can reduce independent challenge. ThoughtSpot lab [researcher inference · strong]

4. Picking

Decision sequence

1. Name the inflection. What changed in architecture, cost, performance, regulation or behavior?
2. Name the constraint. Which workflow/customer pain becomes newly removable?
3. Name the wedge. What narrow product earns urgent adoption before a platform claim?
4. Test founder/team adaptability. What evidence shows learning, recruitment and phase change?
5. Map alternatives. Incumbent bundle, wait, narrower product, adjacent layer or pass.
6. Write dissent and falsifiers. What would show the inflection is real but this company cannot capture it?
7. Freeze terms/vehicle. Price, security, ownership, reserves, adviser and governance rights must fit the evidence stage.
8. Prewrite milestones. Product, customer, GTM, governance and financing gates before reveal or follow-on. [researcher inference · strong]

Red flags and stop conditions

• broad platform language without an urgent first buyer/wedge;
• technology novelty without measured customer value;
• founder purpose that becomes identity defense or withheld information;
• repeat-founder/network familiarity substituting for outside diligence;
• growth/valuation/fundraising substituting for retention, cash controls or governance, as BYJU'S stress-tests;
• follow-on justified by sunk cost rather than a new evidence gate;
• security, reliability, safety or regulatory failure in a trust-critical product;
• financing needs that cannot reach the next credible milestone; and
• a price/ownership path that cannot return the vehicle under a sober high case. Seven decision labs [researcher inference · strong]

5. Deal and portfolio mechanics

The public record does not support a general Mhatre rule for target ownership, check size, security, reserves or concentration. [researcher inference · insufficient public record] Selected filings show substantial firm vehicle positions—Nutanix, MuleSoft, AppDynamics, Rubrik and Netskope—but their cost, distributions and residual chains remain incomplete. Directional ranking

The responsible mechanics are:

• size to evidence, not story intensity;
• model ownership/dilution/down-round/reserve cases before investing;
• treat each follow-on as a new decision with an updated thesis and countercase;
• preserve numbered funds, Select vehicles, SPVs and personal/family holdings separately;
• treat voting/dispositive power as governance, not economics;
• treat issuer valuation/EV/IPO proceeds as company/transaction facts, not investor proceeds; and
• report directional importance until dated cash flows permit MOIC/IRR/DPI/TVPI. [researcher inference · strong]

Portfolio-level check-size distribution, reserve sufficiency, board coverage, hit rate, loss rate and outlier contribution are not computable across 91 associations. Portfolio mechanics gap [researcher inference · insufficient public record]

6. Vehicle and incentive fit

Lightspeed's multi-vintage venture, Select/growth and SPV architecture can support patience and follow-ons. The same architecture creates allocation, governance, duration and attribution questions an individual investor cannot ignore. [researcher inference · strong]

The public record supports distinct LSVP/LSIP/Faction adviser perimeters; multiple UGP directors sharing vehicle power; and separately disclosed Mhatre direct/family positions in Rubrik. It does not disclose fund terms, carry, capital commitments, reserve policy, extensions, recycling, LP liquidity, allocation conflicts or personal compensation. Vehicle economics [documented behavior · strong]

The philosophy fits a long-duration platform with technical specialists, recruiting capacity and follow-on capital. A smaller or reserve-poor vehicle must translate it: narrower sector scope, explicit ownership targets, syndication, milestone-linked reserves and earlier financing-risk kill criteria. [researcher inference · strong]

7. Team attribution and work with founders

The best-supported founder work is phase-specific: help a technical founder turn product into GTM and platform; recruit complementary leadership; use the board to force evidence and adapt to a changed regime. [researcher inference · moderate] Public sources rarely expose which recruit, debate or intervention Mhatre personally owned.

Attribution rules:

• led only where a primary/official source says so (MuleSoft, Rubrik);
• board only with dated issuer/company evidence;
• source, sponsor, diligence, approval, negotiation, board, recruiting, follow-on and exit as separate fields;
• coauthor and visible team credit preserved;
• firm/vehicle outcome never converted into personal return; and
• founder testimony labeled retrospective and relational. Value-add forensics [researcher inference · strong]

8. Views on the craft

• Prepared mind: study technologies deeply enough to recognize an inflection, but do not mistake technical fluency for customer evidence. Stanford, 00:00–08:31 [researcher inference · strong]
• Power law: large outcomes matter, yet missing cash flows and passes make contribution and hit-rate claims unavailable. Portfolio forensics [researcher inference · strong]
• Price: early precision can be false, but price still determines ownership and vehicle return; Mhatre's public record does not justify “price does not matter.” Vehicle/economics controls [researcher inference · strong]
• Board work: useful for recruitment, phase change and governance; the public record seldom reveals the actual intervention. Decision-rights matrix [researcher inference · moderate]
• Exits: an IPO is not realization and an acquisition EV is not investor proceeds; AppDynamics shows public readiness can end in strategic sale. AppDynamics lab [researcher inference · strong]
• Prediction: replace certainty with feedback and regime-aware updating; even direct speech can be factually fallible, as the Stanford science caveat demonstrates. Stanford correction [researcher inference · strong]

9. Signature bets and decision process

• Nutanix tests architecture simplification, staged proof and board duration while preserving the 2010–2011 source conflict and reported-versus-realized mark boundary.
• MuleSoft tests open-source developer adoption becoming an enterprise integration platform; it has the strongest initial/subsequent lead attribution.
• AppDynamics tests founder/product/GTM/platform phase change and a strategic sale that preempted IPO.
• Rubrik tests repeated underwriting, board duration and separate firm/personal security lines.
• Zscaler tests a Series C cloud-security inflection without rewriting it as formation credit.
• ThoughtSpot tests repeat-founder/incubation access, seed/Series A separation and private-valuation bias.
• BYJU'S tests entity, cohort, cash, governance and stop rules without unsupported personal accusation.

Each lab freezes what was known, unknown, believed and disputed; forces an invest/defer/pass choice plus terms, milestones and falsifiers; then reveals the decision/outcome. [researcher inference · strong]

10. Mental models and maxims

1. Detect the technical inflection and name the exact wedge.
2. Hold the long vision; enter through a narrow door.
3. Run lean from zero to one; accelerate only after proof.
4. Back adaptive founders and build the recruiting magnet.
5. Replace static prediction with explicit feedback loops.
6. Re-underwrite when the gravity regime changes.
7. Use the firm platform; keep a case-level attribution ledger.
8. Use repeat relationships for people information; force an outside-view check.
9. Engineer independent readiness; keep IPO and sale optional.
10. Treat trust, safety, reliability and recoverability as adoption infrastructure.
11. Prefer proprietary context, permissions and workflow control over generic model capability.
12. Underwrite geography through local behavior, talent, infrastructure and entity fit.
13. Separate company importance from security price, ownership, vehicle and return.
14. Choose deliberately between an enabling control layer and a full-stack operator.

These are reconstructed teaching rules, not quotations of a hidden Mhatre playbook. Each model records provenance, application, cases, a countercase, exceptions and modern translation. [researcher inference · strong]

11. Evolution and contradictions

12. Failures, misses, and limits

BYJU'S is the strongest adverse lab because the public outcome forces process questions while leaving Mhatre's source, lead, board, stake, loss and causation unresolved. No investor-authored postmortem or documented practice change was found. Failure matrix [researcher inference · strong]

Nutanix derivative litigation was voluntarily dismissed; MuleSoft complaints were allegations disclosed and denied, not misconduct findings. Card anomalies and stale statuses show why official marketing is not an outcome ledger. [contemporaneous record · strong]

The larger limit is missing negative evidence: no complete anti-portfolio, pass, write-off, shutdown or ordinary-outcome corpus. A coherent winner doctrine can be generated by selection and hindsight. Unresolved record [researcher inference · strong]

13. Historical operating environment

Lightspeed's formation crossed the dot-com peak/reset; MuleSoft/AppDynamics began before the GFC; Nutanix/Zscaler grew across the cloud and reopening-IPO era; Rubrik crossed the cheap-capital boom and 2022 reset; 2024–2026 AI deals face large model/capital requirements and fast capability change. Historical context [researcher inference · strong]

Vehicle capacity, public-market windows, incumbent budgets and founder labor markets changed across those regimes. A historical tactic is transferable only after adjusting evidence gates, runway, reserve needs, ownership and exit assumptions. [researcher inference · strong]

SEC filings create valuable dated board/holder snapshots but omit sub-threshold positions, sales/distributions and private economics. Firm cards are more incomplete. Survivor and disclosure bias therefore grow as analysis moves from public headline cases to the 91-card denominator. [researcher inference · strong]

14. Modern VC translation (as of 2026-08-23)

Every translation is dated. It should be rewritten when model economics, regulation, capital costs or exit markets materially change. [researcher inference · strong]

15. Comparison with completed peers

The comparison uses three completed dossiers with different but adjacent methods: David Cowan for enterprise system maps and staged evidence, Danny Rimer for same-vintage global platform/founder service, and Arthur Patterson for prepared-mind category work plus accountable partnership. Documentation depth is not a hit rate, and no row ranks returns. [researcher inference · strong]

The sequential curriculum is Cowan for system maps and explicit evidence purchases; Patterson for preparation, dissent and accountable ownership; Mhatre for inflection/wedge/phase feedback; Rimer for global founder service and cultural contrast. This is a method comparison, not a ranking of alpha. [researcher inference · strong]

16. The strongest case against greatness

Mhatre may be a sophisticated narrator of a selected Lightspeed record rather than a demonstrably superior picker. [researcher inference · strong] Stanford/Bessemer/WPG/Lightspeed created privileged access; multiple partners and founders performed visible work; the official profile exposes association rather than function; winner essays are outcome-aware; BYJU'S is a firm/adviser governance/downside control rather than a clean personal falsifier; and current AI bets remain immature.

No complete opportunity/pass/write-off denominator or security-level cash-flow series exists. Removing any one visible outcome still leaves several company-level successes, but no audited series can prove ex-outlier personal or fund alpha. Skeptical profile [researcher inference · strong]

Original dated memos, full funnel/loss records, partner-function logs, matched ordinary cases and audited cash flows could strengthen or overturn the thesis. Until then the appropriate conclusion is coherent and teachable public decision system; persistent Mhatre-only alpha unproved. [researcher inference · strong]

17. Analyst study guide, glossary, and go deeper

Recommended reading path

1. Writings for dated argument evolution, bylines and selected-success controls.
2. Talks for direct language, transcript provenance and belief changes.
3. Investments for the exact 91-card denominator, economics boundaries and decide-before-reveal labs.
4. Profile for the falsifiable edge and causal alternatives.
5. Mental models for application, countercases and modern translations.

Decision exercises

Complete all seven ex-ante labs without reading the reveal. For each, record invest/defer/pass, terms/ownership, next milestones, falsifiers, dissent and reserve/stop rule.

Compact glossary

Sourcing checklist

• exact source type and first-contact date;
• name/adviser/vehicle/entity and conflict/allocation perimeter;
• technical inflection and customer constraint;
• repeat-relationship/access advantage plus outside-view control;
• source/sponsor/diligence/approval/board owners separated; and
• declined/parked alternatives retained for denominator learning.

Investment and governance checklist

• wedge, buyer, proof, alternatives, dissent and falsifiers;
• check, security, price, ownership, dilution and reserve cases;
• founder learning, complementary recruiting and information behavior;
• reliability/security/safety/regulatory and financing milestones;
• board/information rights and stop/exit conditions; and
• firm, vehicle, team, founder and personal economics kept separate.

Ranked sources to go deeper

1. Stanford full transcript — most complete direct worldview; read it alongside the package's science correction controls, because the transcript contains rather than corrects the two errors.
2. Newcomer full appearance — gravity, feedback and firm history.
3. MuleSoft founder account — strongest personal lead attribution.
4. Nutanix final prospectus — board/vehicle/ownership control.
5. Rubrik 2026 proxy — cleanest firm/personal security separation.
6. BYJU'S failure matrix — adverse/governance boundary and unknowns.

Unresolved research

The highest-value next evidence is the original case memoranda; complete opportunity/pass/write-off ledger; partner-function and board/recruiting logs; term/reserve histories; and audited vehicle cash flows. Without them, do not upgrade directional importance into return or personal-alpha claims. [researcher inference · strong]

Pareto 80/20 — what drove the record

The 80/20 frame prioritizes the few publicly reconstructable company relationships that best explain the visible record; it is not a claim that any measured percentage of Ravi Mhatre's, Lightspeed's, or a specific vehicle's return came from them. No public case supplies a complete initial-and-follow-on cost, dilution, sale/distribution, residual value, fee/carry and dated cash-flow chain, so personal or fund MOIC, IRR, DPI, TVPI, hit rate and loss rate are not computable. [researcher inference · insufficient public record]

The live official profile exposed 91 unique company-card anchors in exact page order at approximately 2026-08-23T13:09Z; its narrative names ten highlights, all subsets of the 91. The card union is a reproducible association denominator—not 91 Mhatre-led investments, 91 board seats, or a return portfolio. [documented behavior · strong] Rain and Raindrop are separate cards; follow-on rounds stay inside one company row.

The most defensible public reconstruction is concentrated in seven enterprise cases—Nutanix, Netskope, MuleSoft, AppDynamics, Rubrik, Zscaler and ThoughtSpot—plus BYJU'S as a genuine adverse/governance test. MuleSoft uniquely has explicit primary wording that Mhatre led the initial and subsequent investment beginning May 2007; Rubrik's current official card also says he led the firm's investment and remained a director. Board service, a firm essay or another profile card does not silently upgrade a case to personal lead. [researcher inference · strong]

Economic-contribution ranking — directional public evidence only

This ranks where the public record provides the strongest combination of disclosed vehicle exposure and reported issuer outcome; it does not rank realized returns, personal economics or causal contribution. [researcher inference · insufficient public record]

Removing any one headline case still leaves other large reported issuer outcomes, but missing cash flows prevent a dominant-outlier or ex-outlier fund-return test. [researcher inference · strong]

Historical-importance ranking

Decision-learning ranking

Denominator, scope and return discipline

Coverage of the reproducible live-card denominator is 91/91; coverage of Mhatre's lifetime opportunity, pass, loss or write-off universe is not computable. [researcher inference · strong] The ledger records Lightspeed page status verbatim or as uncertain—it is not a verified current outcome field, because pages can be stale or internally inconsistent. [documented behavior · strong]

Coverage fraction (Mhatre lifetime opportunity/pass/write-off universe): not computable. The search covered the live profile/card layer; Lightspeed histories, legal/entity pages and writings; company/founder pages; SEC EDGAR issuer/ownership filings; IAPD/ADV; federal-court/docket and adverse routes; archive catalogs; Signal and Form D discovery indexes; named companies, advisers, numbered funds, SPVs and personal/family entities; and the year/decade/name variants Ravi Mhatre, Ravi Brian Mhatre, Ravi LSVP, Mhatre Lightspeed, misspelling Ravi Mahtre, and role/return/complaint terms recorded in the source-map coverage assessment. None exposes the full opportunity/pass/write-off denominator. [researcher inference · strong]

No issuer enterprise value, valuation, IPO gross/net proceeds or acquisition price is called Lightspeed or Mhatre proceeds. No reported firm mark is treated as realized. No shared voting/dispositive position is divided among managers. Later-round subscription slices are never summed as total basis unless the filing proves completeness. [researcher inference · strong]

Historical market, regulatory, and base-rate context

Lightspeed's 1999–2000 formation straddled the dot-com peak and reset. Mhatre later described the crash as a change in operating “gravity,” but the contemporaneous institutional record also shows continuity from WPG Venture Partners and six founding GPs; neither later survival language nor firm marketing is a fund-return series. Light Reading, 2000 · Newcomer, 2023 at 02:01–03:36 [researcher inference · strong]

MuleSoft's 2007 and AppDynamics' 2008 starts preceded the global-financial-crisis financing contraction; Nutanix's 2010–2016 path and Zscaler's 2012 institutional round crossed a long private-company and reopening-IPO cycle. Rubrik's 2014–2024 path crossed the 2020–2021 capital-abundance period and the 2022 reset. Those regimes change follow-on availability, pricing, buyer budgets and exit timing, so reported outcomes cannot be compared without vintage capital and cash-flow data. NVCA/PitchBook Venture Monitor historical series [researcher inference · strong]

Public filings impose useful ownership/board snapshots but threshold-reporting and lockup rules leave sub-threshold positions, later sales, in-kind distributions and personal economics unobserved. Private cards expose still less. The six highlighted enterprise successes are therefore strongly exposed to survivor, disclosure and hindsight bias relative to 91 associations and an unknown pass/write-off universe. [researcher inference · strong]

The 2024–2026 AI cluster appears after rapid frontier-model capability gains and very large private financings. Current valuations and reported revenue/adoption are company or investor claims until independently controlled; the relevant base-rate question is whether present wedges survive model progress and financing normalization, not whether the category is important. [researcher inference · strong]

Vehicle economics and incentives

People and roles map

Comprehensive 91-card ledger

Each row explicitly covers company/profile inclusion; founded and investment date/stage; vehicle, named partner and role; board/ownership; contemporaneous thesis; check/security; follow-ons/dilution; Lightspeed page status and independently verified outcome; reported economics and retrospective account; uncertainty/adverse evidence; and claim-local sources. A dash or missing page field is preserved as unknown, not silently completed. [researcher inference · strong]

Ex-ante decision lab 1 — Nutanix

Exercise status. This is an analyst-created 2010–2011 investment-committee simulation, not a recovered Lightspeed memo. Read through the exercise, make a decision, then reveal the outcome. Where the period record is absent or conflicts, the field says so rather than backfilling it from the IPO. [researcher inference · strong]

Reader exercise — stop before the reveal.

1. Invest, defer or pass, and why?
2. What ownership target and security protections would compensate for the hardware-adjacent capital risk?
3. Which three milestones unlock a follow-on, and which two failures kill it?
4. What evidence would falsify the architecture-simplification thesis before the next round?

Actual decision and reveal. The card records a 2010 Series A, the filing records Mhatre's July 2010 board start, and later Lightspeed vehicles held 27,978,979 shares—22.7% pre/20.3% post and 22.4% voting at IPO. Forbes reported roughly $10 million initially, roughly $40 million total and a roughly $1 billion first-day firm/vehicle mark; those are reported figures, not audited basis, realization or Mhatre economics. Prospectus · Forbes [contemporaneous record · moderate]

Postmortem. The case supports long duration, board continuity and architecture simplification; it does not prove that every original premise survived or that the result was realized. The later derivative action named Mhatre in his board capacity and was voluntarily dismissed, with no misconduct finding. Dismissal stipulation [contemporaneous record · strong]

Ex-ante decision lab 2 — MuleSoft

Exercise status. This analyst-created exercise freezes the information set around May 2007. It is not an IC memo. Founder testimony supplies unusually strong attribution, while original terms, customer diligence and IC mechanics remain unavailable. [researcher inference · strong]

Reader exercise — stop before the reveal.

1. Invest, defer or pass given open-source monetization risk?
2. What security, ownership and governance rights would you request?
3. Which paid-conversion and retention milestones unlock reserves?
4. What evidence would falsify “integration becomes a platform” rather than commodity plumbing?

Actual decision and reveal. Mhatre led the initial and subsequent investment beginning May 2007 and joined the board that month. MuleSoft later completed its IPO; Salesforce agreed to acquire it at approximately $6.5 billion enterprise value and the issuer filed the close. EV is not common-equity proceeds, and neither announcement allocates consideration to Lightspeed or Mhatre. Salesforce agreement · close 8-K [documented behavior · strong]

Postmortem. The strongest learning is explicit early championing plus repeated underwriting; the missing cost/distribution chain prevents a return calculation. Merger complaints were allegations, not findings against Mhatre; the company disclosed and denied the claims, so they belong in governance diligence rather than a misconduct narrative. Company merger supplement · docket [contemporaneous record · strong]

Ex-ante decision lab 3 — AppDynamics

Exercise status. This analyst-created exercise freezes the 2008 Series A/board-start information set. Later founder and investor accounts help reconstruct hypotheses but cannot substitute for the lost original memo. [researcher inference · strong]

Reader exercise — stop before the reveal.

1. Invest, defer or pass at idea stage?
2. How much ownership and what governance rights compensate for single-founder/key-person risk?
3. Which product, GTM and platform milestones govern separate follow-on decisions?
4. What evidence would prove application monitoring is a feature rather than a company?

Actual decision and reveal. Lightspeed invested at Series A and Mhatre joined the board in April 2008. AppDynamics filed to go public, but Cisco acquired it before the IPO for approximately $3.7 billion in cash and assumed equity awards. That transaction value is not Lightspeed proceeds, and the filing does not provide a complete basis or sale allocation. S-1 · Cisco close [contemporaneous record · strong]

Postmortem. The decision lesson is phase-specific underwriting and founder adaptation, not a claim that all phases were predictable in 2008. The strategic sale created an observable public-market-versus-liquidity tension: readiness for IPO did not produce an IPO outcome, and the board's deliberation is not public. [researcher inference · strong]

Ex-ante decision lab 4 — Rubrik

Exercise status. This analyst-created exercise freezes the January 2014/Series A information set. The current company card provides explicit lead/board attribution, but later writings, public-company filings and page/PitchBook metrics must not leak into the initial decision. [researcher inference · strong]

Reader exercise — stop before the reveal.

1. Invest, defer or pass given incumbent and data-loss risk?
2. What ownership, security and board rights do you require?
3. Which reliability, security and GTM milestones unlock a follow-on?
4. What evidence would falsify the platform thesis before breadth consumes the company?

Actual decision and reveal. The official card says Mhatre led the firm investment and remained a director. Multiple Lightspeed vehicles followed; Rubrik later became public. The 2026 proxy's holder lines are dated ownership, not cost or return, and issuer IPO proceeds are not investor proceeds. Mhatre's personal/family lines remain separate. Card · proxy [contemporaneous record · strong]

Postmortem. The case supports repeated underwriting and board duration while exposing why vehicle, personal and issuer economics must remain separate. Page/PitchBook investment figures cannot be reconciled to the proxy without dated cost and security definitions. [researcher inference · strong]

Ex-ante decision lab 5 — Zscaler

Exercise status. This analyst-created exercise freezes the 2012 Series C/first-outside-round information set. Because this was not an idea-stage investment, the decision must be tested as growth/category underwriting rather than rewritten as company formation. [researcher inference · strong]

Reader exercise — stop before the reveal.

1. Invest, defer or pass at Series C rather than company formation?
2. What price/ownership range survives a downside case without knowing the allocation?
3. Which reliability, retention and displacement milestones unlock more capital?
4. What evidence would falsify the cloud-security thesis rather than merely delay it?

Actual decision and reveal. Lightspeed participated in the $38m round and Mhatre articulated the category thesis. Zscaler later completed an IPO, but Lightspeed/Mhatre was not among disclosed 5%+ holders in the inspected filing. No cost, ownership, sale/distribution or personal-economics chain is public. Round release · S-1/A [contemporaneous record · strong]

Postmortem. The case supports correct category direction and later-stage participation, not early formation credit or a measured return. The missing allocation and threshold-limited holder table are the central economics controls. [researcher inference · strong]

Ex-ante decision lab 6 — ThoughtSpot

Exercise status. This analyst-created exercise freezes the 2012 seed/company-formation information set and separately recognizes the issuer's February 5, 2014 $10.7m Series A. The two events cannot be collapsed, and neither source is a recovered memo. [researcher inference · strong]

Reader exercise — stop before the reveal.

1. Invest, defer or pass on a repeat-founder, pre-product seed?
2. What ownership, security and founder-governance protections would you request?
3. Which accuracy, deployment and usage milestones unlock Series A/follow-on capital?
4. What evidence would falsify “search becomes the BI interface” rather than a demo feature?

Actual decision and reveal. The card records the 2012 Seed; the issuer records a firm-led $10.7m Series A in 2014 and Mhatre/Arif board service. ThoughtSpot remained private in the frozen 2026 card; a 2021 Series F led by March Capital at a reported $4.2b valuation is financing/valuation evidence, not Lightspeed proceeds or realized return. Series F release · current board [contemporaneous record · strong]

Postmortem. ThoughtSpot best tests network access and incubation rather than outcome economics. The public record supports company/board continuity but no original ownership, cost, dilution, liquidity or Mhatre-specific lead. [researcher inference · strong]

Ex-ante decision lab 7 — BYJU'S adverse case

Exercise status. This analyst-created exercise freezes the 2016 Series C/global-expansion information set and then reveals the later insolvency process. The official card associates Mhatre with both LSVP and legally distinct LSIP events; it does not prove sole lead, board role, check authority, stake, loss, causation or misconduct. [researcher inference · strong]

Reader exercise — stop before the reveal.

1. Invest, defer or pass at Series C with incomplete cohort and entity evidence?
2. Which adviser/vehicle, security protections and information rights are mandatory?
3. Which learning, cash and governance milestones unlock any follow-on?
4. What evidence would falsify the global-platform thesis before valuation becomes the scorecard?

Actual decision and reveal. The card records both LSVP and LSIP Series C events and associates Mhatre; it does not disclose his decision right. BYJU'S later entered a publicly documented corporate-insolvency-resolution process, and the Supreme Court addressed procedural issues in 2024. Those records do not find wrongdoing by Mhatre or establish his board role, stake, loss or causal responsibility. IBBI process/orders · Supreme Court judgment [contemporaneous record · strong]

Postmortem. This is the strongest adverse control because it turns the playbook toward entity mapping, cohort/cash verification and governance—not because it supports a personal accusation. Without cost, ownership, write-down and cash-flow data, even the investment loss is not quantifiable. [researcher inference · strong]

BYJU'S explicit failure-postmortem record

Decision-rights and value-add forensics

This matrix prevents board presence, firm ownership and later narration from collapsing into a single “led the deal” label. “Unknown” means no inspected public record answers the question; it is not a negative finding. [researcher inference · strong]

Portfolio forensics

Misses, adverse cases and exclusions

Unresolved economics and retrieval record

• No public source supplies a complete initial/follow-on cost, security, dilution, sales/distributions, residual, fee/carry and dated cash-flow chain for any case. Fund/personal MOIC, IRR, DPI, TVPI and realized contribution therefore remain unavailable. [researcher inference · insufficient public record]
• The 91 live cards cover current profile associations, not Mhatre's lifetime deals, opportunities, passes, write-offs, board seats or personal economics. Frozen coverage is 91/91 card-level and 7/91 deep-case; deep unresolved is 84/91 and economic unresolved is 91/91. [researcher inference · strong]
• Original IC memos, diligence work, reference calls, votes, dissent, term sheets, reserve schedules, board books/minutes, recruiting attribution, exit deliberations, LP letters and audited fund cash flows were not located. [researcher inference · insufficient public record]
• Current official cards, SEC/issuer filings, founder testimony, direct writings/talks and adverse searches were reconciled. Third-party databases and firm marketing remain discovery/interested evidence, not silent economic authority. [researcher inference · strong]
• A future upgrade should obtain fund/vehicle cash-flow evidence and contemporaneous early-round documents before changing any directional ranking into a return ranking. [researcher inference · strong]

Coverage and evidence base

Literal priority-source reconciliation

The WRITINGS inventory contains 40 selected priority works: 8 ★★★ and 32 ★★. This is an accessible, decision-relevant corpus, not an exhaustive lifetime bibliography: Medium's author/archive routes are intermittent, several essays migrated or mirror an original, and Lightspeed's archive is dynamic. Each underlying work is counted once. Every processed work receives one note and one substantive subsection here; access, byline and team boundaries are stated per work.

[researcher inference · strong] Evidence boundary. These essays reveal Mhatre's public vocabulary for technical discontinuities, founder ambition, company construction, follow-on conviction, global scaling and platform expansion. They do not disclose a complete opportunity set, passes, IC votes, checks, ownership, reserves, losses, carry, cash flows or counterfactuals. Deal essays are promotional and success-selected; firm/team statements are not person-level performance evidence.

Source-by-source analysis

1. “If Every Company Is a Tech Company, What Do Tech Companies Do?” (2016)

[contemporaneous record · strong] Published 2016-10-25 on Lightspeed's Medium publication under Ravi Mhatre's sole visible byline; the full indexed body was accessible, with named headings and distinctive examples used as locators. byline, date and body

Context, argument, and reasoning. Mhatre argues that digital capability had become mandatory across industries, creating two startup paths: sell infrastructure to incumbents undergoing transformation, or enter an incumbent's domain as a full-stack operator. The core mechanism is data combined with domain expertise. A full-stack entrant must show a product, model or network advantage an incumbent cannot readily copy. GE's sensor-and-operating-knowledge example is illustrative, not comparative evidence. [investor-stated · strong] for the two-path framework; [investor-stated · provisional] for the broad data claim. Essay, “Digital arms dealers” and “Redefining the boundaries of tech”

Decision implications and boundary. Identify enabling layer versus domain operator, then test transformation budget, data rights, domain learning, distribution and incumbent response. The company examples have no failed-company denominator. Later writings narrow the 2016 “data advantage” into trusted context, architecture and production reliability rather than volume alone. [researcher inference · moderate] Essay, closing · source note

2. “What Do You Call a Tech Unicorn That Goes Public?” (2016)

[contemporaneous record · strong] Published 2016-09-19 on Lightspeed's Medium publication under Mhatre's sole visible byline and identifying a TechCrunch original; the full body was accessible and the mirror family counts once, with named headings as locators. byline, date and publication note

Context, argument, and reasoning. In a weak IPO year, Mhatre rejects a special technology “window.” Public investors compare technology with all investable companies, so readiness requires a valuable product, sustained growth, real economics, a delivery organization and customer attachment. Dot-com and unicorn cycles show that technological importance can coexist with overpricing. [investor-stated · strong] for the checklist; [investor-stated · moderate] for the macro account. Essay, “Are tech IPOs coming back?”–“The tech IPO is dead”

Decision implications and tension. Build for durable independent operation; treat flotation as validation rather than strategy. AppDynamics creates a useful tension: a company prepared to remain independent can still accept a strategic acquisition. Neither essay gives a valuation threshold or proves Mhatre controlled an exit. [researcher inference · strong] for the synthesis; [researcher inference · insufficient public record] for causal attribution. IPO essay, closing · AppDynamics essay, acquisition section · source note

3. “IPOs Head to the Enterprise Cloud” (2016)

[contemporaneous record · strong] Published 2016-12-20 on Lightspeed's Medium publication under Mhatre's sole visible byline and noting a LinkedIn original; the full body was accessible, with its numbered three-part framework and named-company paragraphs as locators. byline, date and body

Context, argument, and reasoning. Mhatre forecasts a “componentized enterprise”: legacy applications are recast as services, orchestrated and run on controlled enterprise-cloud infrastructure instead of being wholly replaced by a hyperscaler. Technical modularity is tied to organizational agility and preservation of accumulated processes. Selected vendors illustrate the map but do not prove category durability. [investor-stated · strong] for the three-layer mechanism; [investor-stated · provisional] for the 2017 forecast. Essay, opening and numbered framework

Decision implications and boundary. Locate the indispensable control point, compatibility burden, security requirement and bundling risk. The map transfers as a migration hypothesis, not a license to fund every component. Published 2016-12-20; the source map is corrected. [researcher inference · moderate] Essay, examples and conclusion · source note

4. “Purpose Built: Nutanix” (2016)

[contemporaneous record · strong] Published 2016-09-30 on Lightspeed's Medium publication under Mhatre's sole visible byline; the full IPO-day essay was accessible and its named company-building headings provide locators. byline, date and body

Context, argument, and reasoning. On Nutanix's IPO date, Mhatre argues durable public companies are designed from inception, not timed to a window. His system combines a hard technical problem, platform breadth, a large transition market, early global selling and a team mixing experienced operators with hungry builders. Reported 37% international revenue illustrates global reach but cannot isolate causality. [investor-stated · strong] for the model; [contemporaneous record · moderate] for the metric. Essay, opening through “Mixing experience with entrepreneurialism”

Decision implications and attribution. Make ambition concrete in product breadth, market, international capacity and talent composition, accepting deliberate cost where leadership requires it. “Controlled recklessness” needs demand and operating evidence. The retrospective is promotional; filings control board, vehicle and economics. [researcher inference · strong] Essay, “An international mindset” and closing · source note

5. “Building Jobs, Not Walls” (2016)

[contemporaneous record · strong] Published on Medium 2016-11-24 under Mhatre's sole visible byline and identifying the 2016-11-22 TechCrunch original; the full body was accessible and this underlying work counts once, with narrative/topic paragraphs as locators. byline, date and mirror note

Context, argument, and reasoning. Mhatre connects immigration policy to his father's path from immigrant engineer to operator of a distressed startup and to two decades of immigrant-founded technology companies. His mechanism is greenfield creation: successful new markets create products, exports and job ecosystems. The company list and cited unicorn study are examples, not causal policy evaluation. [investor-stated · strong] for biography and position; [investor-stated · provisional] for aggregate effects. Essay, family account, company list and greenfield argument

Decision implications and boundary. The biography helps explain a recurring preference for immigrant, globally ambitious technical founders and market creation over incumbent protection. It does not establish a picking advantage, and its “color-blind” framing understates barriers the essay does not examine. Medium mirrors the TechCrunch original and counts once. [researcher inference · moderate] Essay, closing · source note

6. “Building a Multi-Billion Company the AppDynamics Way” (2017)

[contemporaneous record · strong] Published 2017-01-25 on Lightspeed's Medium publication under Mhatre's sole visible byline; the full post-offer retrospective was accessible and its two “Moral” headings and acquisition section provide locators. byline, date and body

Context, argument, and reasoning. Mhatre describes two-stage construction: founders with “extreme purpose” anticipated fragmented cloud applications and paired end-to-end monitoring with consumer-grade enterprise delivery; later leadership added operating scale without discarding product velocity. Small initial deployment with Global 2000 scalability preserved expansion. [investor-stated · strong] for the model; [researcher inference · moderate] for causal success attribution because the piece followed Cisco's offer. Essay, opening, vision and “Moral…Part I”

Decision implications and exit tension. Pick founder foresight plus a product wedge that expands into a platform; scale through complementary leadership rather than founder mythology. Mhatre says the company intended to float but accepted Cisco's proposal, framing acquisition strength as the result of not needing a sale. The public record does not expose price deliberation or his authority. [researcher inference · strong] Essay, acquisition and “Moral…Part II” · source note

7. “Doubling Down — Our Investment in Rubrik” (2017)

[contemporaneous record · strong] Published 2017-05-09 on Lightspeed's Medium publication under Mhatre's sole visible byline; the full body was accessible, comprising his introduction and a speaker-labeled founder interview used as locators. byline, date and Q&A

Context, argument, and reasoning. Mhatre's introduction and interview document close formation: Bipul Sinha had been a Lightspeed partner, Rubrik incubated in its office for eight months, and recruiting was directly observable. The follow-on case combines founder insight, legacy data-protection pain, cloud-native architecture, early adoption and recruiting. The reported $100m run rate and $140m cumulative commitment are company/firm claims, not audited cash flows or personal economics. [investor-stated · strong] for relationship and stated diligence; [investor-stated · provisional] for metrics. Essay, introduction and Q&A

Decision implications and attribution. Proximity can improve observation but creates network-selection and promotional risks. Test a follow-on on independent customers, product velocity, management depth and reserve concentration. “Our” means Lightspeed; filings control vehicle and personal holdings. [researcher inference · strong] Essay, early-customer and vision answers · source note

8. “Predicting Enterprise Trends in 2018” (2018)

[contemporaneous record · strong] Published 2018-01-16 on Lightspeed's Medium publication under Mhatre's sole visible byline; the full forecast was accessible and its four named trend headings provide locators. byline, date and body

Context, argument, and reasoning. Mhatre forecasts hybrid cloud; vertical AI/data; blockchain relevance despite speculation; enduring cyber demand; and computing moving into medicine, vehicles and physical systems. Across themes the mechanism is orchestration of heterogeneous infrastructure and data, while global connectivity increases security demand. He admits founders' envisioned futures sometimes fail. [investor-stated · strong] for the dated thesis; [investor-stated · provisional] for forecasts. Essay, opening and four trend sections

Decision implications and forecast audit. Test whether an investable control point, team and economics emerged instead of declaring broad themes “correct.” Hybrid orchestration, security and physical computing persisted; the claim that cryptocurrency might rival the internet lacked a token/company/value-capture test. [researcher inference · strong] Essay, cryptocurrency and closing · source note

9. “India's Mobile Internet Revolution and ShareChat's $100M Series C” (2018, with Dev Khare)

[contemporaneous record · strong] Published 2018-09-27 on Lightspeed's Medium publication with Ravi Mhatre and Dev Khare as visible coauthors; the full body was accessible, with market, thesis and round-chronology paragraphs used as locators. byline, date and body

Context, argument, and reasoning. Mhatre and Khare argue cheap mobile data and vernacular participation were expanding India's internet beyond English-speaking urban users. ShareChat's language-first product, iteration, engagement/retention and low-friction nonpaid growth are selection signals. Lightspeed first invested at Series A and joined the Series C; the essay does not allocate leadership or economics between authors, LSVP and legally distinct LSIP. [investor-stated · strong] for the thesis and attribution boundary. Essay, market context through round chronology

Decision implications and boundary. Observe behavior at the edge of a distribution-cost discontinuity: retention, organic acquisition, language density and repeated learning matter more than population statistics. This writing does not prove Mhatre led ShareChat. [researcher inference · strong] Essay, product evidence and round paragraph · source note

10. “Zscaler: Innovation and Audacity Win the Race” (2018)

[contemporaneous record · strong] Published 2018-03-16 on Lightspeed's Medium publication under Mhatre's sole visible byline; the full retrospective was accessible, with distinctive architecture, engineering, scale and investment-history paragraphs as locators. byline, date and body

Context, argument, and reasoning. Mhatre reconstructs a contrarian 2008 security thesis: as mobile and cloud dissolved the perimeter, a cloud service between user and destination could replace appliance-heavy security. He pairs Jay Chaudhry's serial-founder record with a difficult requirement—inspection at high throughput without material latency—and later reported footprint and customer metrics. [investor-stated · strong] for the architecture; [investor-stated · provisional] for selected metrics. Essay, founder background through scale metrics

Decision implications and attribution. Separate prediction from ability to make the future usable at scale. The essay says Lightspeed joined the first outside financing in 2012 and followed in 2015; it discloses no allocation, personal ownership or return. IPO validates public access, not counterfactual superiority. [researcher inference · strong] Essay, engineering problem and closing investment history · source note

11. “The Future of Business Intelligence” (2019, with Amy Wu)

[contemporaneous record · strong] Published 2019-08-28 on Lightspeed's Medium publication: structured metadata names Mhatre, while the body is signed Ravi Mhatre and Amy Wu. The full body was accessible, with problem, company-history, metrics and lifecycle paragraphs as locators. byline, signature, date and body

Context, argument, and reasoning. The authors frame BI as a human bottleneck: more than a billion knowledge workers rely on roughly 1.5m specialist analysts, so questions take days or weeks. ThoughtSpot's search interface, real-time visualization and machine learning are offered as a way to move inquiry to the business user without sacrificing enterprise security. Repeat-founder evidence is central: Lightspeed knew Ajeet Singh from Nutanix, helped incubate ThoughtSpot and led its Series A, then led a reported $248m Series E. Customer, growth and run-rate figures are company/firm claims. [investor-stated · strong] for the product mechanism; [investor-stated · provisional] for metrics. problem, product, history and metrics paragraphs

Decision implications, assumptions and evolution. A repeat-founder network can reduce team uncertainty and enable incubation, but it can also create privileged-access and familiarity bias. The later 2021 Glean essay and 2023 Glean essay update this self-service-search thesis from analytics to permission-aware enterprise knowledge and generative answers. Actual leadership and economics remain team/vehicle facts; the co-signed piece does not prove Mhatre alone led each round. [researcher inference · strong] ThoughtSpot history and lifecycle paragraph · source note

12. “Announcing the Lightspeed Enterprise CEO Advisory Council” (2019)

[contemporaneous record · strong] Published 2019-10-02 on Lightspeed's Medium publication under Mhatre's sole visible byline and signature. The full body was accessible; council-purpose and four member-history paragraphs provide locators. byline, date and body

Context, argument, and reasoning. Mhatre formalizes an existing enterprise network into an advisory council for operations, board construction, mentorship and partnerships. The founding members—David Wadhwani, Greg Schott, John W. Thompson and Bipul Sinha—are linked to AppDynamics, MuleSoft, Rubrik and other portfolio work. The mechanism is reusable operating knowledge and executive matching, not brand alone; named recruiting histories make the intervention more concrete, although the essay supplies no measured effect. [documented behavior · strong] for formation and named roles; [investor-stated · provisional] for category-winner causality. council-purpose and member-history paragraphs

Decision implications, assumptions and boundary. The platform model assumes experienced executives' knowledge transfers across companies and that advice does not blur governance or conflicts. Diligence should ask which founders used the council, under what authority, with what counterfactual outcome. This is firm-platform evidence, not Mhatre's personal board track record or proof the council caused exits. [researcher inference · strong] purpose and closing · source note

13. “Next-Level Defense vs Next-Gen Security Threats / Confluera” (2019, with Ansaf Kareem)

[contemporaneous record · strong] Published 2019-07-30 on Lightspeed's Medium publication; structured metadata names Mhatre while the body is signed Ravi Mhatre and Ansaf Kareem. The full body was accessible, with threat, product, round and team paragraphs as locators. byline, signature, date and body

Context, argument, and reasoning. The essay argues porous cloud/mobile perimeters, stealthy lateral movement, alert overload and scarce security skill make post-mortem tooling insufficient. Confluera's proposed wedge is real-time attack tracking and surgical response across cloud/hybrid environments, while contextualizing third-party alerts. Lightspeed says it led the Series A and credits founders Abhijit Ghosh, Niloy Mukherjee and Bipul Sinha plus participating executives. Early-customer resonance is asserted without names or denominator. [investor-stated · strong] for the problem/product mapping; [investor-stated · provisional] for customer validation. threat, product and round paragraphs

Decision implications, assumptions and boundary. Underwrite whether real-time context improves precision, detection and response without becoming another alert layer; test integration, false positives and incumbent response. This extends Zscaler's perimeter-displacement thesis into SecOps workflow, but the signed coauthor and named Sinha roles require shared credit. No outcome evidence shows Mhatre personally caused the investment or operational result. [researcher inference · strong] product and team paragraphs · source note

14. “Aurora: Powering the Next Revolution in Transportation” (2019, with Raviraj Jain)

[contemporaneous record · strong] Published 2019-02-08 on Lightspeed's Medium publication; metadata names Mhatre and the body is signed Ravi Mhatre and Raviraj Jain. The full body was accessible, with problem, technical-complexity, founder and round paragraphs as locators. byline, signature, date and body

Context, argument, and reasoning. The authors connect autonomy to road safety, accessibility, time and transportation-as-a-service, then emphasize that computer vision, sensor fusion, mapping, ML and robotics require multidisciplinary technical leadership. Aurora's founders are selected for prior Google/Waymo, Uber/CMU and Tesla experience; reported iteration speed and humility supplement credentials. Lightspeed joined a financing alongside Sequoia, Greylock and Index, but the piece does not say it led. [investor-stated · strong] for team/problem logic; [investor-stated · provisional] for speed and market-impact claims. technical-complexity, founder and financing paragraphs

Decision implications, assumptions and evolution. For physical-world systems, pedigree must translate into safe engineering, iteration evidence, capital endurance and platform extensibility. The 2021 IPO essay adds partnerships and sensing architecture, but neither proves safety or economic deployment. Co-signature and firm “we” prevent solo attribution; later public filings govern ownership and outcome. [researcher inference · strong] founder, challenge and closing physical-world paragraphs · source note

15. “Growing Our Team at Lightspeed” (2020)

[contemporaneous record · strong] Published 2020-07-21 on Lightspeed's Medium publication under Mhatre's sole visible byline. The full institutional essay was accessible; its four named personnel paragraphs provide locators. byline, date and body

Context, argument, and reasoning. Mhatre presents diversity of viewpoint, sector specialization, growth-stage experience and executive recruiting as components of firm value-add. Mark Anderson brings operating/board experience; Shan Shan public-market and consumer/software analysis; Nnamdi Iregbulem technical enterprise and programming; Pete Steinle executive recruiting and operating experience. The mechanism is a deliberately heterogeneous platform around founders, not a single generalist partner. [documented behavior · strong] for hires and roles; [investor-stated · moderate] for expected value-add. opening and four biographies

Decision implications, assumptions and boundary. A platform should be tested on decision rights, handoffs, founder utilization and whether specialization improves outcomes net of coordination cost. This evolves the 2019 CEO Council from external operating advisers to internal investment/talent capacity. The announcement contains no outcome denominator and cannot be used as Mhatre-only doctrine or performance. [researcher inference · strong] opening, roles and closing · source note

16. “Enabling a Blameless Culture” (2019)

[contemporaneous record · strong] Published 2019-03-20—not 2020—on Lightspeed's Medium publication under Mhatre's sole visible byline. The full body was accessible, with problem, SRE mechanism, relationship, seed/Series A and adoption paragraphs as locators. byline, date and body

Context, argument, and reasoning. Mhatre argues incident response fails when organizations search for a person to blame rather than root causes and prevention. Blameless operationalizes the alternative through planned error budgets and reliability orchestration across product, development and operations. He credits founders' firsthand SRE pain and network history through Box, MuleSoft and Lightspeed; the firm co-led seed and Series A after reported launch/adoption progress. Named customers are selected testimonials, not a cohort. [investor-stated · strong] for the organizational mechanism; [investor-stated · provisional] for adoption/outcome claims. opening, platform, founder and financing paragraphs

Decision implications, assumptions and boundary. The transferable rule is to measure learning, prevention and cross-team coordination rather than blame avoidance as a slogan; error budgets must not excuse poor reliability. The follow-on illustrates conviction after product and customer progress, but the essay lacks reserve thresholds, independent results or failed comparisons. [researcher inference · strong] platform and seed-to-Series-A paragraphs · source note

17. “Materialize: Streaming Without Compromises” (2020, with Raviraj Jain and Nnamdi Iregbulem)

[contemporaneous record · strong] Published 2020-11-30 on Lightspeed's Medium publication; metadata names Mhatre while the body credits Ravi Mhatre, Raviraj Jain and Nnamdi Iregbulem. The full body was accessible, with trade-off bullets, architecture, team and financing paragraphs as locators. byline, body credit, date and body

Context, argument, and reasoning. The authors define a database trade-off: batch is stale, existing streaming is limited/expensive, and Lambda combines complexity. Materialize recasts SQL queries as incremental dataflows over live data, preserving PostgreSQL compatibility and consistency while avoiding repeated recomputation. Founder technical experience and the open-source Timely Dataflow engine support feasibility. Lightspeed led the Series A and participated in a Kleiner Perkins-led Series B. [investor-stated · strong] for the technical mechanism and disclosed round roles; [investor-stated · provisional] for “without compromises.” trade-off, architecture, team and financing paragraphs

Decision implications, assumptions and boundary. Technical diligence should ask which compromises merely moved to cost, operations, workload fit or market adoption. Familiar interfaces can reduce migration friction, while a research pedigree is evidence only when production benchmarks and customers validate it. Three-person authorship and firm round language require shared attribution; no economics or causal board intervention is public. [researcher inference · strong] architecture, roadmap and financing · source note

18. “YugabyteDB: Distributed SQL Has Arrived” (2021, with Nnamdi Iregbulem)

[contemporaneous record · strong] Published 2021-03-03 on Lightspeed's Medium publication; metadata names Mhatre and the body is signed Ravi Mhatre and Nnamdi Iregbulem. The full body was accessible, with architecture, founder, community, customer-reason and financing paragraphs as locators. byline, signature, date and body

Context, argument, and reasoning. Microservices create demand for resilient, scalable, geo-distributed systems of record. Yugabyte's wedge combines distributed execution with PostgreSQL compatibility, reducing migration friction; founders' Cassandra/HBase and Facebook experience plus reported community/customer usage support technical credibility. Lightspeed led a $48m follow-on after backing the Series A. Community, installed-cluster and benchmark figures are issuer/firm claims. [investor-stated · strong] for the problem/product and round history; [investor-stated · provisional] for scale evidence. opening through customer reasons and financing

Decision implications, assumptions and boundary. Underwrite compatibility, consistency, failover, total operating burden, developer adoption and proof that distributed requirements are broad enough for an independent platform. The essay continues the componentized-cloud thesis with a more exact database wedge. Co-signature prevents sole credit, and the post-follow-on essay has no pass/loss denominator or personal economics. [researcher inference · strong] architecture, customer-reason bullets and financing · source note

19. “Dexterity: Ushering a New Era of Intelligent Robots” (2020, led by Raviraj Jain)

[contemporaneous record · strong] Published 2020-07-21 on Lightspeed's Medium publication under Raviraj Jain's lead/structured byline and signed Raviraj Jain and Ravi Mhatre. The full body was accessible, with labor, product/RaaS, founder and seed-history paragraphs as locators; this is a signed team work, not a Ravi-led article. lead byline, signature, date and body

Context, argument, and reasoning. The article connects warehouse labor attrition and e-commerce capacity growth to demand for flexible automation. Dexterity's wedge is hardware-agnostic intelligent software on commodity robots, sold as robot-as-a-service to reduce customer capital and time to ROI. The team says it led the 2017 seed after judging Samir Menon's cross-functional ability, ambition and later customer delivery. Labor, construction, market and performance figures are promotional/selective. [investor-stated · strong] for the stated mechanism; [investor-stated · provisional] for scale and “uncapped” market claims. labor, technology/RaaS and founder paragraphs

Decision implications, assumptions and boundary. Physical automation diligence should test throughput, exception rate, safety, integration, utilization and customer payback—not only technical demos. RaaS shifts adoption friction but places capital/reliability risk on the provider. Lead authorship belongs to Jain; Mhatre's signature supports team association but not personal sourcing, lead or return. [researcher inference · strong] technology/RaaS and seed-history paragraphs · source note

20. “The Convergence of Technology and Health” (2021, with Will Kohler and Jonathan MacQuitty)

[contemporaneous record · strong] Published 2021-06-29 on Lightspeed's official site with a visible body byline for Ravi Mhatre, Will Kohler and Jonathan MacQuitty; structured “Authors” metadata lists only Mhatre. The full body was accessible, with named team and investing sections as locators; the visible body byline governs. date, body byline and full article

Context, argument, and reasoning. The authors frame patient impact as the objective and describe an evolution from genomic diagnostics/platforms into therapeutics, digital care and multi-stage health investing. They give explicit credit to Jonathan MacQuitty, Ling Wong, Galym Imanbayev and Shelley Chu and explain formation-to-growth capacity across three reported fund pools. Ling's role in Guardant and the team's hires are firm/team evidence, not Mhatre-only investments. [documented behavior · strong] for team/practice development; [investor-stated · provisional] for impact and portfolio claims. opening, history, team profiles and “Investing in Health”

Decision implications, assumptions and boundary. The platform thesis is that scientific, clinical, operating and financial expertise must combine across long development cycles, with flexible vehicles supporting formation through public markets. The public essay does not disclose clinical attrition, partner allocation, losses or returns, and the 2021 deployed-capital figure is a firm snapshot. It marks sector expansion rather than proof Mhatre personally underwrote each named company. [researcher inference · strong] team sections, fund paragraph and closing · source note

21. “Better-Lending Platform Blend Goes Public” (2021)

[contemporaneous record · strong] Published 2021-07-16 on Lightspeed's Medium publication under Mhatre's sole visible byline. The full IPO-day body was accessible, with founder-history, Wells Fargo, platform-scale and portfolio paragraphs as locators. byline, date and full body

Context, argument, and reasoning. Mhatre frames Blend as modernization software for banks constrained by legacy systems and slow change. The founder case rests on the four cofounders' Palantir work during the financial crisis; Lightspeed says that confidence drove its 2013 Series A lead. Wells Fargo's later adoption becomes the proof point that an ambitious young team could serve highly regulated institutions, while 31 of the top 100 financial firms and $1.4tn of 2020 applications are selected company/firm metrics. [investor-stated · strong] for the disclosed selection logic and round role; [investor-stated · provisional] for operating scale. founder crucible through customer/scale paragraphs

Decision implications, assumptions and boundary. In slow-moving regulated markets, test whether founder-earned domain knowledge, enterprise trust and a low-friction product can overcome institutional inertia. One marquee client can update conviction but is not a customer cohort. “We” and a firm-wide fintech history do not establish Mhatre's individual sourcing, board authority, economics or causation; the filing governs vehicle ownership. [researcher inference · strong] Series A, Wells Fargo and firm-fintech paragraphs · source note

22. “Offchain Labs: How Arbitrum Is Transforming Ethereum Scalability” (2021, led by Amy Wu)

[contemporaneous record · strong] Published 2021-08-31 on Lightspeed's Medium publication under Amy Wu's structured/lead byline and a visible body credit to Amy Wu and Ravi Mhatre. The full launch/financing article was accessible, with trilemma, product, developer-selection and closing round paragraphs as locators. lead byline, coauthor credit, date and body

Context, argument, and reasoning. The authors map Ethereum's fee/throughput constraint to an optimistic-rollup layer that executes off-chain and posts transaction summaries to Ethereum. Their stated Series B decision gives developer usability priority over theoretical architecture: bytecode compatibility, quick integrations, hundreds of announced partners, delivery and fair access. Lightspeed says it participated in Series A and led the $100m Series B; TVL, fee-reduction and throughput figures were contemporaneous ecosystem/company claims. [investor-stated · strong] for the decision criteria and disclosed round roles; [investor-stated · provisional] for market and performance figures. trilemma, developer-choice and financing paragraphs

Decision implications, assumptions and boundary. For protocol infrastructure, compare real developer migration cost, compatibility, reliability and credible neutrality—not only theoretical elegance or token activity. The essay's billion-user outlook lacked a monetization, security-incident or bear-market test, so the dated portfolio argument cannot validate a durable outcome. Wu is lead author; Mhatre's co-credit supports team association, not sole sourcing, leadership, ownership or return. [researcher inference · strong] developer-choice and closing paragraphs · source note

23. “Glean: Bringing Search to the Workplace” (2021, with Ansaf Kareem)

[contemporaneous record · strong] Published 2021-09-15 on Lightspeed's Medium publication; structured metadata names Mhatre and the visible body credit names Ravi Mhatre and Ansaf Kareem. The full launch essay was accessible, with SaaS-fragmentation, founder-history, product, metrics and financing paragraphs as locators. byline, coauthor credit, date and body

Context, argument, and reasoning. SaaS proliferation creates fragmented permissions, content and context; Glean's horizontal assistant indexes applications and models language, behavior and workplace relationships. The team case is unusually relationship-rich: Lightspeed knew Arvind Jain through Riverbed and Rubrik, while his Google search experience fit the exact problem. Lightspeed says it co-led the initial funding and notes a General Catalyst-led follow-on; 40 customers and two hours saved weekly are company/firm claims. [investor-stated · strong] for problem/team/round logic; [investor-stated · provisional] for customer/productivity metrics. problem, founder, product and financing paragraphs

Decision implications, assumptions and boundary. A founder's prior context is strongest when it matches a specific technical and behavioral bottleneck, but familiarity still needs independent product, permission and customer diligence. The 2022 follow-on essay and 2023 generative-AI essay evolve search into trusted knowledge and permission-aware answers. Coauthorship and firm “we” preclude sole personal credit or economics. [researcher inference · strong] founder and platform paragraphs · source note

24. “Driving Innovation: Aurora Goes Public” (2021, led by Raviraj Jain)

[contemporaneous record · strong] Published 2021-11-04 on Lightspeed's Medium publication under Raviraj Jain's structured/lead byline and a visible body credit to Raviraj Jain and Ravi Mhatre. The full SPAC-day body was accessible, with “Exceptional Team,” “Partnerships that Accelerate” and “Best-in-class Technology” headings as locators. lead byline, coauthor credit, date and full body

Context, argument, and reasoning. The authors say Lightspeed surveyed more than 100 autonomy startups and selected Aurora for founder experience, ecosystem partnerships and a multimodal sensing/simulation stack. Aurora planned to start with trucking before passenger mobility; partner and virtual-testing evidence supported commercialization claims. The $9.4tn market, employee, patent and simulated-mile equivalents are issuer/firm claims, while the listing followed a SPAC merger rather than a conventional IPO. [investor-stated · strong] for the stated evaluation dimensions; [investor-stated · provisional] for scale and safety implications. opening and three named sections

Decision implications, assumptions and boundary. Capital-intensive systems require a staged wedge, safety evidence, OEM/logistics distribution and endurance beyond a public transaction. The article advances the 2019 team thesis with observable partnerships and sensing architecture but still does not prove safe driverless unit economics. Lead authorship belongs to Jain; Mhatre's co-credit does not establish personal lead, ownership or result attribution. [researcher inference · strong] commercialization plan and closing · source note

25. “Investing in Alchemy” (2022)

[contemporaneous record · strong] Published 2022-02-08 on Lightspeed's Medium publication under Mhatre's sole visible byline. The full financing essay was accessible, with platform analogy, infrastructure, adoption-loop, team and closing market paragraphs as locators. byline, date and full body

Context, argument, and reasoning. Mhatre argues large computing shifts require developer platforms and casts Alchemy as Web3's AWS-like abstraction: reliable, scalable, accurate blockchain infrastructure plus development tooling. The claimed loop is more builders → applications → users → more builders. Lightspeed says it led a $200m round alongside existing investors; 140% team growth, 6,000% API-call growth, NPS and “majority” usage assertions were promotional, live-market figures near the 2021–22 crypto peak. [investor-stated · strong] for the platform mechanism and round role; [investor-stated · provisional] for market leadership and metrics. platform, infrastructure, loop, growth and financing paragraphs

Decision implications, assumptions and contradiction. Underwrite indispensability across cycles through uptime, data correctness, developer retention, pricing power, protocol concentration and application survival. The Windows/AWS analogy usefully asks who controls the developer layer but risks skipping decentralization and commoditization. A promotional snapshot near a market peak is no proof of a durable return or Mhatre-specific alpha. [researcher inference · strong] adoption loop, growth snapshot and closing forecast · source note

26. “The Future of Enterprise Search: Glean” (2022, with James Ephrati and Ansaf Kareem)

[contemporaneous record · strong] Published 2022-05-18 on Lightspeed's Medium publication; structured metadata names Mhatre while the visible body credit names Ravi Mhatre, James Ephrati and Ansaf Kareem. The full Series C essay was accessible, with product, founder, usage and round-history paragraphs as locators. byline, three-author credit, date and body

Context, argument, and reasoning. The authors refine Glean from federated search to “connective tissue”: unified discovery incorporates activity, related knowledge, verified answers and embedded interfaces. The repeat-founder account again links Arvind Jain to Riverbed, Rubrik and Google. Lightspeed says it co-led the 2019 initial financing, joined the 2021 Series B and invested through the Growth team in the Series C; productivity and marquee-customer claims remain selected. [investor-stated · strong] for lifecycle chronology and product expansion; [investor-stated · provisional] for usage outcomes. product, founder, metrics and round chronology

Decision implications, assumptions and evolution. Follow-on conviction should test whether the wedge expands in function and distribution while permissions and relevance hold at scale. Cross-vehicle lifecycle support can preserve access, but “triple down” discloses neither price, reserves nor independent return. Three-author and Growth-team framing require shared credit; the 2023 essay adds generative answers grounded in trusted knowledge. [researcher inference · strong] feature and financing paragraphs · source note

27. “Building Blockchain Infrastructure, One Startup at a Time: Meet Venture Partner Anand Iyer” (2022)

[contemporaneous record · strong] Published 2022-07-20 on Lightspeed's Medium publication under Mhatre's sole visible byline. The full institutional announcement was accessible, with “Doubling down on crypto” and “An operator and an investor” headings as locators. byline, date and full body

Context, argument, and reasoning. Mhatre presents blockchain infrastructure as an unfinished platform layer and hires Iyer for operator, angel, teaching and infrastructure experience. The essay names more than ten Lightspeed investors and a joint venture with independently managed Faction, while reporting $600m firm-wide blockchain deployment within $1.5bn fintech deployment. It also lists FTX in the portfolio. [documented behavior · strong] for the hire, team and disclosed firm figures; [investor-stated · provisional] for paradigm and market claims. opening, “Doubling down on crypto,” and biography

Decision implications, assumptions and adverse boundary. Specialized platform hiring can deepen sourcing and diligence, but decision rights, entity structure and correlated exposure must be explicit. FTX's inclusion fixes one public firm association, but the essay supplies no Mhatre role, diligence or outcome and the portfolio list is not validation. Firm deployment figures, joint-venture networks and named investors are not Mhatre's personal portfolio, ownership or return. [researcher inference · strong] crypto-team and portfolio paragraphs · source note

28. “Lightspeed Adds Experienced Growth Investor to the Team” (2018, multi-signed)

[contemporaneous record · strong] Published 2018-07-10 on Lightspeed's Medium publication under Mhatre's structured byline, but the body is signed Will Kohler, Natalie Luu, Ravi Mhatre, Jeremy Liew and Team Lightspeed. The full announcement was accessible, with fund-family, growth-mandate and Twohig biography paragraphs as locators. metadata byline, date, multi-party signature and body

Context, argument, and reasoning. The signatories describe Lightspeed as early-stage at its core while reserving $1.05bn of a reported $1.8bn fund family for follow-ons and selected new growth companies with product-market fit, executive depth and scalability. Brad Twohig's cross-stage and cross-geography experience is hired to support the lifecycle model. Success counts are biographical claims without audited attribution or loss denominator. [documented behavior · strong] for funds, mandate and hire; [investor-stated · provisional] for performance and platform benefits. fund-family, mandate and biography paragraphs

Decision implications, assumptions and boundary. Stage-specific talent and vehicles can help retain access as companies mature, but later rounds need price, ownership and cross-fund-conflict discipline rather than automatic support. This prefigures the 2022 Glean lifecycle narrative and 2021 health fund flexibility. The signature makes this institutional doctrine, not a Mhatre-only decision or personal AUM/return claim. [researcher inference · strong] growth mandate and closing · source note

29. “The Future of Higher Education: Multiverse” (2022, led by Nakul Mandan)

[contemporaneous record · strong] Published 2022-06-08 on Lightspeed's official site; the body credit names Nakul Mandan, James Ephrati, Nicole Quinn and Ravi Mhatre, while structured “Authors” metadata lists only Mandan. The full body was accessible under named headings; the visible four-person body credit governs. date, body credit, structured-author boundary and article

Context, argument, and reasoning. The authors frame Multiverse as an employer-paid apprenticeship marketplace and training platform that substitutes paid applied learning for costly college. Lightspeed says it led the seed, remained a major investor and Growth co-led the $220m Series D. Retention, promotion, diversity and customer figures are company/firm claims; the marketplace may improve access but the essay does not compare matched college outcomes, long-term wages or unit economics. [investor-stated · strong] for model and round chronology; [investor-stated · provisional] for social and operating outcomes. opening, “How…work?”, metrics, “Why…matter?” and financing

Decision implications, assumptions and boundary. For education/labor marketplaces, test employer willingness to pay, apprentice completion, wage mobility, credential portability and marketplace liquidity—not mission language alone. The multi-stage firm commitment is not evidence Mhatre led any round. Mandan is lead/structured author; Ephrati, Quinn and Mhatre share body credit, so personal sourcing, governance and economics remain unproven. [researcher inference · strong] model, outcomes and closing financing · source note

30. “Typeface: Transforming Enterprise Content Creation with Radically Personalized Generative AI” (2023, with Raviraj Jain and company/AI assistance)

[contemporaneous record · strong] Published 2023-02-27 on Lightspeed's Medium publication under Mhatre's structured byline and signed Ravi Mhatre and Raviraj Jain. The full body was accessible under “A world-class team…” and “Generative AI…” headings. A closing disclosure says Lightspeed and Typeface co-wrote it and used Typeface to brainstorm, create images and craft full sentences; the work is direct evidence of published positioning, not unaided personal prose. byline, signature, date, body and disclosure

Context, argument, and reasoning. The piece identifies enterprise content velocity, personalization and brand control as the wedge between general-purpose generation and workplace adoption. Founder Abhay Parasnis's Adobe/cloud/AI background and reported customer interest support team/timing; Lightspeed announces participation in a $65m round. The essay forecasts competitive advantage but offers no accuracy, copyright, security, retention or revenue evidence. [investor-stated · strong] for the stated wedge and investment; [investor-stated · provisional] for differentiation and demand. problem/product, team and investment paragraphs

Decision implications, assumptions and contradiction. Enterprise generative AI diligence should test proprietary context, brand controls, provenance, workflow fit, measurable output gains and model dependence. The co-creation disclosure is unusually useful: it demonstrates adoption while weakening any attempt to treat every sentence as Mhatre's independent language. Joint signature, company collaboration and AI assistance require shared authorship; firm investment does not expose lead, price or return. [researcher inference · strong] enterprise-use and disclosure paragraphs · source note

31. “Glean's Enterprise-Ready Generative AI Is Transforming Workplace Search and Discovery” (2023)

[contemporaneous record · strong] Medium metadata dates the underlying Mhatre sole-byline work 2023-05-01, while the current Lightspeed migration displays 2023-05-02; the full official body was accessible under “A foundation of trusted knowledge” and “A critical head-start.” The migration-date discrepancy is preserved rather than flattened. current author, date and full article · Medium archive route

Context, argument, and reasoning. General-purpose models lack current internal facts, relationships, terminology and permissions. Mhatre argues that Glean's four years building enterprise search created a trusted knowledge model and ranking foundation on which generative answers, expert detection and contextual recommendations can operate. The repeated Jain relationship and 2019 initial co-lead connect founder pattern recognition to product architecture; capability and “ideal solution” claims remain promotional. [investor-stated · strong] for the context/permission mechanism; [investor-stated · provisional] for product superiority. opening, trusted-knowledge, product and head-start sections

Decision implications, assumptions and evolution. Enterprise AI needs current grounded context, access control, retrieval quality and provenance—not just a strong base model. This narrows the 2016 “data advantage” and 2021–2022 search thesis into an operational trust layer. No benchmark, security audit, error cohort, price or Mhatre-specific economics is disclosed. [researcher inference · strong] trusted-knowledge and feature sections · source note

32. “Deepening Our Partnership with Anthropic” (2025, with Guru Chahal and Sebastian Duesterhoeft)

[contemporaneous record · strong] Published 2025-09-02 on Lightspeed's official site with an Authors block naming Ravi Mhatre, Guru Chahal and Sebastian Duesterhoeft. The full Series F essay was accessible under “Singular Focus,” “When Focus Meets Opportunity” and “Our Partnership.” date, authors and body

Context, argument, and reasoning. The authors distinguish AI from earlier platform shifts because scaling model capability purportedly creates new abilities. Anthropic's narrow mission around trustworthy/safe intelligence, talent and enterprise applicability forms the selection case; three consecutive Lightspeed investments express follow-on conviction. Reported $1bn-to-$5bn run-rate growth, autonomous-coding hours and Claude Code revenue are issuer/firm claims, not audited results or realized value. [investor-stated · strong] for the stated thesis and investment chronology; [investor-stated · provisional] for growth and capability metrics. three named sections

Decision implications, assumptions and boundary. For frontier labs, test whether mission focus, capability progress, safety work, distribution and capital endurance compound rather than merely consume capital. The essay assumes model intelligence retains value over applications and that safety supports enterprise adoption; the separate AI-safety essay makes that mechanism explicit. Three authors and “Lightspeed” prevent sole credit, and Series F participation discloses no personal allocation, price, ownership or return. [researcher inference · strong] partnership and closing · source note

33. “Doubling Down on Lightspeed's Investment in Descope” (2025, with Guru Chahal)

[contemporaneous record · strong] Published 2025-09-30 on Lightspeed's official site with an Authors block naming Guru Chahal and Ravi Mhatre. The full seed-extension essay was accessible under “The Trusted Identity Layer” and “Looking Ahead.” date, authors and full body

Context, argument, and reasoning. AI agents, MCP servers and machine workflows expand identity from human login toward permissions between machines. Descope's developer-first, no/low-code authentication layer and the Demisto team's shared history form the product/team case. Lightspeed participated in a seed extension led by GGV/Notable; customer, revenue and 75–80% go-live-speed claims are company/firm figures without a disclosed comparison protocol. [investor-stated · strong] for the problem/product and round role; [investor-stated · provisional] for traction and performance. opening, trusted-layer and financing paragraphs

Decision implications, assumptions and boundary. Identity may become a control point for agentic systems if the product handles authorization, federation, misuse and developer integration better than incumbents. Test active production use, security outcomes, switching friction and expansion from consumer to agent identity. Chahal/Mhatre coauthorship and a firm participation—not lead—prevent solo decision or return attribution. [researcher inference · strong] trusted-layer, evolution and financing sections · source note

34. “Investing in Unconventional AI: Biology-Scale Efficiency for the AI Era” (2025, with Guru Chahal and Bucky Moore)

[contemporaneous record · strong] Published 2025-12-08 on Lightspeed's official site with an Authors block naming Guru Chahal, Ravi Mhatre and Bucky Moore. The full investment essay was accessible as a compact narrative from energy constraint through architecture, founders and round, followed by current legal/third-party disclaimers. date, authors, article and disclaimers

Context, argument, and reasoning. The authors frame energy and cost—not demand—as the AI-compute bottleneck, contrasting kilowatt-scale GPUs with the brain and the theoretical Landauer limit. Unconventional proposes computing neural-network dynamics through physics rather than conventional digital simulation. Cross-disciplinary founders and homogeneous AI workloads support the timing case; Lightspeed says it co-led/led the round alongside Andreessen Horowitz, while power projections and efficiency gaps rely on unverified third-party/company claims. [investor-stated · strong] for the stated architecture and round; [investor-stated · provisional] for constraint timing and achievable efficiency. energy, architecture, team and financing paragraphs

Decision implications, assumptions and boundary. Deep-compute bets require staged physics validation, manufacturability, software/toolchain fit, yield, latency and application economics before extrapolating theoretical headroom. A large energy need may justify R&D but not select this implementation. Three authors and current firm disclaimers preserve team, source and live-mark boundaries; no benchmark, ownership or realized return is public. [researcher inference · strong] architecture, financing and disclaimers · source note

35. “AI Safety vs AI Scalability Is a False Dichotomy” (2025)

[contemporaneous record · strong] Published 2025-10-27 on Lightspeed's official site under Mhatre's sole Authors entry. The full thesis was accessible under “Cybersecurity's Lesson,” “The Gaps Are Already Here,” “The Market Is Bifurcating,” “Anthropic Proves the Thesis” and “Safety Scales.” date, author and full body

Context, argument, and reasoning. Mhatre argues that enterprise markets bifurcate around trust: built-in reliability, auditability, misuse prevention, steerability and defense enable rather than slow deployment. He uses cybersecurity and Rubrik as analogy, then Anthropic contract/revenue claims as a case that capability plus safety wins regulated/high-stakes buyers. The commercial mechanism is coherent; causal language, selected examples, “fastest” claims and several technical-risk assertions are not independently demonstrated by the essay. [investor-stated · strong] for the safety-as-enabler framework; [investor-stated · provisional] for causal and market-validation claims. cybersecurity, gaps, bifurcation and Anthropic sections

Decision implications, assumptions and tension. Underwrite safety as measurable product architecture: evaluations, audit trails, red-team performance, incident response, model control and customer procurement evidence. Reliability may accelerate regulated adoption, but uniform standards and firm portfolio examples do not prove which regime or company wins. This gives a 2025 trust constraint to the 2018 broad technology-inflection outlook. [researcher inference · strong] requirements, market and closing sections · source note

36. “Backing America's AI Leadership: Applauding the White House AI Action Plan” (2025, firm-authored with a direct Mhatre quotation)

[contemporaneous record · strong] Published 2025-07-23 on Lightspeed's official site as a firm-policy essay without a personal Authors block; one block quote is explicitly attributed to Mhatre. The full page was accessible as a narrative policy statement. Only that quote is direct Mhatre voice; the remainder is firm-authored positioning. date, firm text and direct quotation

Context, argument, and reasoning. The firm supports a national AI framework, streamlined regulation, infrastructure/energy buildout and investment across frontier models, infrastructure and applications. Mhatre's direct claim is that owning the frontier prevents commoditization and that the plan supports scaling and energy supply. This is a policy preference tied to portfolio exposure; it does not evaluate alternative regulatory, safety, environmental or geopolitical outcomes. [investor-stated · strong] for Mhatre's quoted position; [documented behavior · strong] for firm portfolio framing; [investor-stated · provisional] for policy causality. quotation, regulatory, energy and stack paragraphs

Decision implications, assumptions and boundary. Policy and energy are underwriting variables for capital-intensive AI, but national-competitiveness rhetoric should not substitute for company-specific moat, price or downside analysis. Regulatory simplification can accelerate entrants while reducing safeguards; abundant energy has timing and externality constraints. Firm authorship means the page supports one Mhatre maxim, not his authorship of every claim. [researcher inference · strong] quote, policy and infrastructure paragraphs · source note

37. “Our Investment in Fireworks AI: the Inference Platform Aiming to Power Every GenAI Application” (2025, five authors)

[contemporaneous record · strong] Published 2025-10-28 on Lightspeed's official site with an Authors block naming Anoushka Vaswani, Ravi Mhatre, Sebastian Duesterhoeft, Guru Chahal and Justin Shen. The full Series C essay was accessible under “The Fireworks Mission,” “From Zero to 10 Trillion Tokens Daily” and “Our Partnership,” followed by current disclaimers. date, five authors, body and disclaimers

Context, argument, and reasoning. Fireworks is framed as an inference/control layer that lets developers customize open models, manage compute and optimize cost, latency and quality without a closed API dependency. Customer conversations, the ex-PyTorch team and production focus drive the case; Lightspeed says it led the Series C alongside Index and Evantic. Customer, token, speed and cost metrics are firm/company claims without benchmark methodology. [investor-stated · strong] for the selection logic and round role; [investor-stated · provisional] for performance and category-leadership claims. mission, diligence, metrics, team and financing

Decision implications, assumptions and boundary. Inference infrastructure diligence should compare workload quality, latency, utilization, reliability, gross margin, model portability and retention under falling compute prices. Developer control may create value, but hyperscaler bundling and model commoditization could compress it. Five-author credit and firm disclaimers make this team evidence; no personal lead, ownership or return allocation is public. [researcher inference · strong] mission, metric and partnership sections · source note

38. “Investing in Ricursive Intelligence: AI for Chip Design and Chip Design for AI” (2026, with Guru Chahal and Jonah Cader)

[contemporaneous record · strong] Published 2026-01-26 on Lightspeed's official site with an Authors block naming Guru Chahal, Ravi Mhatre and Jonah Cader. The full Series A essay was accessible as a narrative from design bottleneck through recursive architecture, prior work, team and financing. date, authors and full body

Context, argument, and reasoning. Manual chip design takes large teams, years and substantial cost while software advances faster. Ricursive proposes a full-stack design system and recursive loop: AI improves workload-specific silicon, which accelerates the next design cycle. Founders Goldie and Mirhoseini's Google work and reported MediaTek adoption support technical credibility; Lightspeed says it led the Series A. Time/cost and adoption claims are selected, and full-chip closure remains unproven publicly. [investor-stated · strong] for bottleneck, mechanism and round; [investor-stated · provisional] for compression and adoption. problem, architecture, prior-work and financing paragraphs

Decision implications, assumptions and boundary. Test whether AI-generated designs meet physical constraints, verification, manufacturability, portability and customer integration—not just subproblems or research benchmarks. A feedback loop is investable only if each cycle yields defensible data/product gains. Three authors require shared credit; no check allocation, board role, price, ownership or outcome is disclosed. [researcher inference · strong] architecture, prior-work and closing · source note

39. “Announcing Our Investment in Temporal: Agents Make Mistakes. Your Workflows Can't.” (2026, four authors)

[contemporaneous record · strong] Published 2026-02-17 on Lightspeed's official site with an Authors block naming Anoushka Vaswani, Ravi Mhatre, Bucky Moore and Justin Shen. The full Series D essay was accessible under “The Pattern We Kept Seeing,” “Why Temporal Wins,” “Temporal's Traction” and “Our Partnership,” followed by current disclaimers. date, authors, body and disclaimers

Context, argument, and reasoning. Hundreds of stated founder/customer conversations revealed a repeated gap between agent demos and long-running production: retries duplicate work, state disappears and errors compound. Temporal abstracts event histories, recovery and state so ordinary code resumes after failure; its distributed-systems history is reframed as agent infrastructure. Lightspeed announces a $300m Series D; growth, installs and customer usage are company/firm claims. [investor-stated · strong] for observed pattern and technical mechanism; [investor-stated · provisional] for category leadership and traction. pattern, product, traction and financing sections

Decision implications, assumptions and boundary. Look for repeated failure evidence across customers, then test recovery correctness, developer experience, workflow lock-in, observability and economics. An old infrastructure primitive can gain a new wedge without becoming an AI-native monopoly. Four authors and firm diligence preclude Mhatre-only sourcing or outcome attribution; “hundreds” lacks a sampling record. [researcher inference · strong] pattern, product and partnership sections · source note

40. “Move 37, Ten Years On: Partnering with Ineffable Intelligence” (2026, with Raviraj Jain and Antoine Moyroud)

[contemporaneous record · strong] Published 2026-04-27 on Lightspeed's official site with an Authors block naming Ravi Mhatre, Raviraj Jain and Antoine Moyroud. The full seed essay was accessible under “Distillation and discovery,” “What Ineffable is building” and “Building this, safely.” date, authors and full body

Context, argument, and reasoning. The authors distinguish distilling finite human data from experiential systems that generate data through interaction. David Silver's progression from DQN/AlphaGo through MuZero/AlphaProof supplies a prior-work arc; the proposed system would act in engineering environments and learn from feedback. Lightspeed says it co-led the seed and cites research pedigree, team magnetism and safety disposition. Those signals support a frontier bet, not proof the system will discover useful knowledge. [investor-stated · strong] for the learning paradigm, prior-work case and round; [investor-stated · provisional] for uniqueness and scientific promise. opening, three named sections and financing

Decision implications, assumptions and evolution. Frontier research diligence can use a decade of progressively harder prior work, research-talent attraction, testable environments and a builder's safety disposition. It must still define reward validity, transfer from games/formal domains, compute cost, control and milestone falsifiers. This extends the 2018 physical-world computing thesis into self-generated knowledge while retaining the 2025 safety constraint. Three authors and co-lead language require shared institutional credit. [researcher inference · strong] research arc, experiential-learning and safety sections · source note

Cross-corpus synthesis

Core argument and what the writings can prove

[researcher inference · strong] Across the selected 2016–2026 corpus, Mhatre's most consistent public argument is not “pick a large market.” It is: identify a technical or distribution discontinuity; locate the exact bottleneck whose removal changes user behavior; back a team with both unusual purpose and relevant earned insight; enter through a narrow product that can widen into a platform; and build the organization, capital base and trust required to survive a long journey. The sequence is visible from the two-path technology-company framework, through AppDynamics' end-to-end wedge, Glean's trusted context, and Temporal's durable execution.

[researcher inference · strong] The corpus supports a public decision vocabulary and a record of disclosed firm actions. It cannot prove personal alpha, causal value-add or returns. Deal essays are promotional and success-selected; follow-on posts omit price and reserve thresholds; coauthors and firm vehicles often own the action; private opportunity sets, passes, write-offs, IC votes and cash flows are missing. The strongest conclusions are therefore about stated process and repeated observable patterns, not performance.

Dated chronology and evolution

Decision playbook

Recurring decision rules and hidden assumptions

1. Start with the discontinuity, then name the indispensable wedge. [researcher inference · strong] Componentized cloud, perimeterless security, streaming SQL, distributed databases, inference and durable execution all move from a broad shift to a specific bottleneck. The hidden assumption is that the wedge is important enough to support an independent platform and not merely a hyperscaler feature. enterprise cloud · Zscaler · Temporal
2. Pair a long vision with a narrow first proof. [researcher inference · strong] AppDynamics' initial deployment, Aurora's trucking entry, Glean's search bar and Descope's customer identity are wedges toward wider platforms. The assumption is that adjacent expansion preserves product quality and economics. AppDynamics · Aurora · Descope
3. Treat founder evidence as adaptive capacity, not résumé prestige. [researcher inference · strong] Repeat work, crisis-earned domain context, product judgment, delivery history and talent attraction recur across Blend, Glean and Move 37. The assumption—and risk—is that privileged network access predicts future adaptation rather than familiarity bias. Blend · Glean · Move 37
4. Update conviction from observable proof, but separate proof from promotion. [researcher inference · strong] Rubrik proximity, Blameless adoption, Glean product expansion and Fireworks customer conversations exemplify follow-on updates. The missing assumptions are valuation, reserve opportunity cost and counterfactual customer cohorts. Rubrik · Blameless · Fireworks
5. Build for independence even when an acquisition is rational. [researcher inference · strong] Public readiness and AppDynamics are compatible only if readiness is an option-generating operating standard, not a command to float. The public record does not reveal the reservation price or decision authority. IPO essay · AppDynamics acquisition account
6. Make trust, safety and reliability part of the product. [researcher inference · strong] Blameless, Glean, AI safety and Temporal turn operational trust into adoption infrastructure. The hidden assumption is that buyers will pay for architecture rather than accept a bundled or retrofitted alternative. Blameless · Glean GenAI · AI safety
7. Use a firm platform, but preserve ownership of the decision. [researcher inference · strong] Specialist investors, advisers, growth vehicles and cross-stage support can improve company service. They also increase handoff, conflict and attribution risks. Named coauthors, round leads, entities and vehicle facts must remain attached. Growth-team hire · health platform · crypto platform

Dated contradictions and belief updates

Strongest countercontrols

• Public independence versus strategic sale. [researcher inference · strong] The 2016 readiness doctrine says durable operation comes first; the 2017 AppDynamics essay celebrates a sale immediately before a planned IPO. The reconciled rule is optionality, but price and authority remain inaccessible. 2016 · 2017
• Broad forecasts versus investable specificity. [researcher inference · strong] The 2018 outlook's hybrid/security themes translate into exact later wedges; its suggestion that cryptocurrency could rival the internet lacked a company/token/value-capture test. The 2022 Alchemy and crypto-team pieces sharpen infrastructure selection but remain cycle-peak, promotional evidence. 2018 forecast · Alchemy
• Data advantage versus trusted context. [researcher inference · strong] The 2016 formulation is broad; by 2023 the doctrine requires current context, permissions, ranking and provenance. This is an update, not confirmation that raw data volume is a moat. 2016 · 2023
• Speed at an inflection versus trust before deployment. [researcher inference · strong] Earlier essays reward rapid iteration and boldness; 2025–26 writings insist capable systems remain prototypes without reliability and control. The updated rule is fast learning inside explicit safety/recovery constraints. enterprise trends · AI safety · Temporal
• Repeat-founder access versus denominator bias. [researcher inference · strong] Riverbed/Rubrik/Glean and Nutanix/ThoughtSpot relationships can improve observation, yet the corpus does not show comparable familiar-founder passes or failures. Network access is a sourcing advantage claim, not a proven selection advantage. ThoughtSpot · Glean
• Capital continuity versus price discipline. [researcher inference · strong] “Double/triple down” and lifecycle vehicles communicate support but never reveal price, reserves, dilution, cross-fund conflicts or foregone alternatives. Follow-on frequency is not evidence of follow-on quality. Growth mandate · Glean Series C · Anthropic Series F
• Direct voice versus mediated publication. [contemporaneous record · strong] Twenty-three works have visible coauthors/multi-signatures, sixteen appear sole-authored, and one is institutional prose with only a direct Mhatre quote. Typeface additionally discloses company co-writing and AI-generated sentences. Structured metadata repeatedly omits body coauthors (Blameless, health, Multiverse), so visible body credit governs. Typeface disclosure · Blameless migration · Multiverse

Authorship, team, vehicle and decision attribution

[contemporaneous record · strong] The 40-work denominator contains 16 sole-author presentations, 23 visible coauthored/multi-signed works, and one institutional policy page with a direct Mhatre quotation. “Sole author” describes public presentation, not proof that no editor or company contributed. Coauthored articles establish shared thesis publication; a firm “we” establishes a Lightspeed claim; neither alone establishes who sourced, led an IC vote, occupied a board seat or received economics.

[researcher inference · strong] A safe attribution ladder is: visible author/signatory → explicitly named partner role → explicitly disclosed firm round role → filing-proven vehicle/board fact → only then person-level decision or economics if a primary record says so. Examples: Amy Wu leads the Offchain article; Raviraj Jain leads both Aurora/Dexterity articles; Guru Chahal shares Descope/Unconventional work; five partners share Fireworks; Mhatre's quote does not make him author of the AI-policy page. Shared beneficial voting/dispositive power, firm AUM, round size and company exit value are not personal cost, proceeds or return.

Modern VC translation (as of 2026-08-23)

Five best works and unresolved record

Ranked five for decision insight

This ranks decision-process usefulness, not literary merit, deal return or personal ownership.

1. “Building a Multi-Billion Company the AppDynamics Way” (2017). [researcher inference · strong] Best combination of technical inflection, product wedge, enterprise delivery, complementary leadership and a real exit-option tension; it also exposes how retrospective success can overstate causality. essay
2. “Doubling Down — Our Investment in Rubrik” (2017). [researcher inference · strong] Best observable account of incubation, relationship-based diligence, recruiting evidence and follow-on conviction, with a clean warning that proximity and firm commitment are not personal economics. essay/Q&A
3. “Glean: Bringing Search to the Workplace” (2021). [researcher inference · strong] Strongest ex-ante mapping of a repeat founder's exact experience to a fragmented-data/permission problem, an initial product and disclosed funding roles; later essays permit a real evolution audit. essay
4. “Purpose Built: Nutanix” (2016). [researcher inference · strong] Clearest compact company-construction doctrine: hard problem, platform ambition, large transition, global early, and mixed-experience talent—balanced by IPO-day selection bias. essay
5. “Move 37, Ten Years On” (2026). [researcher inference · strong] Best current frontier-research memo: prior-work trajectory, learning mechanism, talent signal, safety disposition and explicit long horizon, while leaving scientific transfer and economics honestly unproven. essay

Access, exact denominator and unresolved retrieval gaps

[contemporaneous record · strong] Literal coverage arithmetic: 40 selected (8 ★★★, 32 ★★); 40 full-text; 0 partial; 0 inaccessible/metadata-only; 40 deeply analyzed; 40 notes. This is an accessible priority corpus, not an exhaustive lifetime bibliography. Medium's author/archive pages are intermittent; current Lightspeed migrations can change dates, structured Authors blocks and canonical URLs; mirrors/newsletters count once at underlying-work level. The Glean 2023 one-day date discrepancy, Blameless/health/Multiverse structured-metadata omissions and Typeface's co-writing/AI disclosure are preserved rather than silently normalized.

[researcher inference · strong] Remaining writing-specific gaps are exact and material: no full historical export of Mhatre's Medium account; no private drafts or editorial record; no proof that “sole author” means unaided authorship; no complete map of removed/migrated posts; no IC memos, board materials, LP letters or private shareholder communications; no pass/write-off denominator; no valuation/reserve/ownership history for follow-ons; and no audited link between any essay's thesis and personal cash-flow performance. Those limits block an exhaustive bibliography, authorship forensics and a personal-return score, but not the bounded process analysis above.

Coverage and evidence base

The bounded public-web corpus contains 31 distinct appearances: 24 YouTube appearance families plus seven non-overlapping external readouts. The YouTube side began with 25 substantive source assets; the two complementary IIT Bay Area 2017 videos belong to one event family, yielding 24 appearances. Twenty-one families expose provider captions. Three did not; their audio was acquired only after caption and official-transcript routes failed, then transcribed one at a time through the OpenAI API. Six of seven external appearances expose usable edited text or notes. The Bloomberg Nutanix 2016 page exposes only title, date, speaker and topic metadata. Final reconciliation: 31 selected; 30 text-usable; one partial/metadata-only; 31 attempted and analyzed; 31 notes created. Mirrors, excerpts and alternate players remain routes, not extra appearances.

Source-by-source analysis

1. The Language of Technology — Stanford eCorner (2020)

Publication and access. Stanford published the 45:10 speaker-series video on 2020-11-05 and a seven-page official timestamped transcript on 2020-11-11. The video and official transcript are one underlying appearance. Mhatre is the interview subject; Ravi Belani is host. Provider captions and official PDF timestamps were cross-checked. See source note.

This is the richest first-person process source in the bounded corpus. [researcher inference · strong] At 05:46, Mhatre says venture requires continuous listening and learning; at 06:28, he treats every interaction as a relationship rather than a transaction. [investor-stated · strong] The actual-decision translation is to make founder learning velocity and relationship durability observable, then keep separate proof for deal role, vehicle and outcome. [researcher inference · strong]

The talk is also a warning about direct-speech fallibility. [researcher inference · strong] The official transcript's life-sciences discussion on p. 6 records Mhatre's “4.3 billion” human-DNA-pair figure at 40:05 and implication that an AIDS vaccine was developed at 40:52. [investor-stated · strong] Those premises are not reliable: NHGRI says the human genome contains approximately three billion base pairs, and HIV.gov says no vaccine is currently available to prevent or treat HIV. [researcher inference · strong] The statements evidence only his 2020 life-sciences interest and direct-speech fallibility, not scientific facts. [researcher inference · strong]

2. Venture Capital Gravitational Physics — Newcomer (2023)

Publication and access. Eric Newcomer published the 43:15 interview on 2023-03-29; the captioned video and publisher page are one appearance. Newcomer asks the questions; Mhatre answers. Provider automatic captions are full but machine-generated. See source note.

Mhatre describes a reset as changed “rules of gravity”: prior assumptions about capital, product and buyers stop compiling (02:51–03:36). [investor-stated · strong] He recalls that Lightspeed entered the dot-com crash shortly after raising its first fund and retained an early-stage, high-conviction objective (02:01–02:38). [investor-stated · moderate] This is retrospective firm history, not audited performance. [researcher inference · strong]

The actual-decision rule is regime-sensitive underwriting: re-test burn, financing availability, customer budgets and time-to-proof whenever gravity changes. [researcher inference · strong] The contradiction is useful: the metaphor rejects static prediction, yet the interview is itself success-selected and cannot reveal the companies that failed to recompile. [researcher inference · strong]

3. Lightspeed Cofounders' 17-Year Overnight Success — StrictlyVC (2017)

Publication and access. The 15:02 panel was published 2017-05-10 by StrictlyVC/TechCrunch. The captioned video and edited TechCrunch transcript are one appearance. Connie Loizos interviews Barry Eggers, Mhatre and Bipul Shah; claims are speaker-mapped. See source note.

Mhatre says the enterprise-investor founders first established Series A enterprise credibility before expanding the platform (01:47–02:24). [investor-stated · strong] He says AppDynamics was essentially one founder when Lightspeed decided to fund it, then Greylock joined (05:16–05:47). [investor-stated · strong] The latter is decision-process evidence, while filings and contemporary witnesses govern dates and shared lead credit. [researcher inference · strong]

At 10:54–11:31, Mhatre argues venture returns compress as excess capital enters and prices disconnect from public markets. [investor-stated · moderate] It is a thesis, not a demonstrated causal estimate, and it creates a dated tension with later mega-fund platform expansion. [researcher inference · strong]

4. Nutanix at .NEXT / theCUBE (2015)

Publication and access. SiliconANGLE/theCUBE uploaded the 23:34 interview from Nutanix .NEXT on 2015-06-09. The video has full provider captions. Dave Vellante and Stu Miniman host; Mhatre is the interview subject. Captions misrecognize names, so company spellings were normalized only where the speaker context is unambiguous. See source note.

Mhatre says the founding team's boldness mattered because it joined deep stack complexity with a simple user experience (02:12–03:36). [investor-stated · strong] He frames hybrid/cloud infrastructure transition as a five-to-ten-year migration rather than a single product cycle (07:15–08:00). [investor-stated · moderate]

The actual-decision rule is long vision plus narrow architectural wedge: require a technical reason incumbents cannot cheaply retrofit, while testing whether simplification changes adoption. [researcher inference · strong] Because this is a company-sponsored conference and pre-IPO success narrative, it cannot establish return or isolate Mhatre's contribution. [researcher inference · strong]

5. TiEcon keynote interview (2017)

Publication and access. TiE Silicon Valley uploaded the 22:48 interview on 2017-05-16. The video has full provider captions. The interviewer prompts Mhatre; caption speaker labels are absent, so claims are assigned only when the turn is clear. See source note.

Mhatre recounts choosing in 1999 between operating and starting a venture firm with three college contacts; he emphasizes cofounder trust over certainty about the business (02:29–04:08). [investor-stated · strong] This is his four-person formation memory and must coexist with the contemporaneous six-GP/WPG institutional record. [researcher inference · strong] He also recalls the first crash, impaired investments and the need to return capital to earn another fund (08:05–08:21). [investor-stated · moderate] No amount or fund-return proof is supplied. [researcher inference · strong]

His founder rule combines disruptive-product ambition with endurance through periods when the outcome is unknown (06:51–08:46). [investor-stated · strong] Modern use requires milestones that distinguish perseverance from unproductive escalation. [researcher inference · strong]

6. Snorkel AI fireside/product session (2022)

Publication and access. Snorkel AI uploaded the 1:01:46 session on 2022-04-27. The video has full provider captions. CEO Alex Ratner delivers the product presentation and later converses with Mhatre; most early technical claims are Ratner's, not Mhatre's. See source note.

Ratner describes Snorkel's research-to-company path and a programmatic data-development loop (04:01–12:33). [contemporaneous record · moderate] These are founder/company claims and therefore context for, not statements by, Mhatre. The investment-relevant observation is the conversion of a research bottleneck into an iterative product workflow; customer lists and speed claims remain unverified promotional evidence. [researcher inference · moderate]

This appearance is retained because it reveals what the investor chose to platform publicly, but the actual deal rationale is thinner than in Mhatre-led interviews. [documented behavior · moderate] It should not be used to attribute Snorkel technical assertions or operating results to him. [researcher inference · strong]

7. Hill & Valley Forum — AI and the American Fabric (2025)

Publication and access. The 26:55 video was published 2025-05-03; the YouTube video and Apple podcast are mirrors of one panel. Mhatre moderates Senator Mike Rounds and Anthropic cofounder Jack Clark. Provider captions are full. See source note.

Mhatre introduces Lightspeed as lead investor in Anthropic's reported $3.5 billion financing, then directs questions on safety, China, export controls and open models (01:21–02:10). [investor-stated · moderate] Financing is firm/team attribution, not personal capital or return. [researcher inference · strong] Substantive regulatory positions in the session belong mainly to Rounds and Clark—for example Clark's testing-regime argument at 17:07–17:58—and are not Mhatre claims. [contemporaneous record · strong]

The documented behavior is agenda selection: Mhatre uses portfolio access to convene policymakers and a portfolio-company executive around innovation/safety tradeoffs. [documented behavior · strong] Sponsorship, moderator status and investment interest make this a conflicted policy setting; it is not neutral expert evidence. [researcher inference · strong]

8. MuleSoft founder–investor fireside (2018)

Publication and access. IIT Bombay Heritage Foundation uploaded a 34:40 captioned excerpt on 2018-09-07; it is part of the longer LM7kZsSRNBM event, so both count once. The captioned excerpt pairs MuleSoft CEO Greg Schott with Mhatre; a host introduces them. See source note.

Schott, not Mhatre, supplies much of the operating narrative. He describes product-market fit as harder to observe in enterprise software and says culture/team questions were initially underweighted (06:31–11:55). [retrospective witness · strong] Later he emphasizes customer-first decisions and hiring (13:23–16:24). [retrospective witness · strong]

The source supports a founder/investor learning model, not sole-investor attribution. Mhatre's board/investment role must be cross-checked to Ross Mason's account and the S-1, while Schott's statements remain his own. [researcher inference · strong]

9. AppDynamics founder journey — Startup Grind (2018)

Publication and access. Startup Grind uploaded the 19:45 Jyoti Bansal/Mhatre interview on 2018-06-04. The video exposes no provider caption track and no official transcript was located. After those routes failed, audio was acquired to temporary RAM storage and transcribed 2026-08-23 with OpenAI whisper-1 verbose segment timestamps; propositions were reviewed, and neither media nor full transcript enters Git. See source note.

Bansal describes three company phases: product-market fit, repeatable go-to-market, then addressable-market and platform expansion (01:40–03:35). [retrospective witness · strong] Bansal says the founder must evolve from storyteller to product leader and later organizational builder (06:06–08:45). [retrospective witness · strong] The joint lesson is phase-specific founder adaptation; it is still a post-exit success retrospective. [researcher inference · strong]

10. Bloomberg Technology — the future of crypto (2021)

Publication and access. Bloomberg uploaded the 4:55 interview on 2021-12-10. The video has no provider captions or official transcript. Temporary audio was transcribed 2026-08-23 through OpenAI whisper-1 with segment timestamps after caption routes failed; no full transcript/media is committed. See source note.

Mhatre argues both under- and over-regulation create risk (00:06–01:08) and says Lightspeed had invested in crypto since 2013 across several market cycles (03:39–04:38). [investor-stated · moderate] The second is a firm-wide history, not a personal ledger. [researcher inference · strong] His long-run bullishness is a dated forecast; later crypto failures and regulatory events are essential counterevidence. [researcher inference · strong]

11. NYSE Interesting Times, Episode 4 (2024)

Publication and access. NYSE's canonical 7:30 upload is dated 2024-09-10; i85E4mJuWYE is a 2024-10-01 same-channel duplicate. The canonical video has no caption track or official transcript. Temporary audio was transcribed 2026-08-23 using OpenAI whisper-1 segment timestamps; no full transcript or media is committed. See source note.

Mhatre marks ChatGPT as a broad realization that computing could automate parts of cognition (00:28–01:30). [investor-stated · strong] He warns that anthropomorphic interfaces, fake-person scale and security create new risks, then says investors have a responsibility to fund safeguards (02:30–04:02). [investor-stated · strong] His near-term AGI inevitability is a belief, not established fact. [researcher inference · strong]

12. TiEcon interview (2012)

Publication and access. TiE uploaded the 14:32 interview on 2012-10-03; a shorter BPcxfbezsvA excerpt is a duplicate. The captioned full video is the canonical route. The interviewer and Mhatre alternate; captions are automatic. See source note.

Mhatre says he entered venture in 1995 after seeing Netscape-era company formation and wanting exposure to multiple new companies rather than one operator role (04:01–05:45). [investor-stated · strong] His father's scientist-to-CEO path is presented as formative evidence for technical entrepreneurship (01:40–03:32). [investor-stated · moderate] Family testimony is direct but not independently verified operating history. [researcher inference · strong]

13. IIT Bay Area 2017 appearance family

Publication and access. The event family has two complementary captioned assets: a 41:06 fireside BF9UGJHjP7k uploaded 2017-06-22 and a 22:06 conference/community video F7q5U4gQBV8 uploaded 2017-07-29. Provider metadata and content show different edits from one event; they count once. The second asset is a multi-speaker montage but contains a unique, unambiguous Mhatre window at 18:14–20:33. See source note.

In the fireside, Mhatre gives a detailed family-mobility account: his father moved from biochemistry research into business leadership and a Stanford spinout, influencing Mhatre's technology/company lens (02:13–05:20). [investor-stated · moderate] In the complementary edit, he connects that example to early-stage work, Bessemer entry, starting a firm with Stanford classmates and a technical-founder wedge (18:14–20:33). [investor-stated · strong] His “all four of us” core-team memory must coexist with the contemporaneous WPG/Lightspeed record naming six founding GPs. [contemporaneous record · strong]

14. CNBC-TV18 — India investment interview (2015)

Publication and access. CNBC-TV18 uploaded the 12:44 interview on 2015-09-26. The video has provider automatic captions. The host asks; Mhatre answers. See source note.

Mhatre says his professional India-office work began in 2005 and argues experience recycling plus smartphone/data adoption were maturing the ecosystem (01:23–02:41). [investor-stated · strong] He describes India as more domestically driven than export-led and, at that moment, attractive on valuation/upside (02:55–04:30). [investor-stated · moderate] This is a 2015 forecast, not proof of later India-vehicle performance, and LSVP/LSIP attribution must remain separate. [researcher inference · strong]

15. Stanford ZKP/blockchain panel (2024)

Publication and access. Stanford's 51:10 course panel was uploaded 2024-07-12. The video has a manual en-US track and automatic captions. Mhatre opens with his history, then moderates cryptography founders/researchers; panelist claims remain theirs. See source note.

Mhatre says Lightspeed's first blockchain investment was in 2013 and frames the thesis as a distributed-computing architectural shift whose initial digital-money use could pull a broader platform into existence (01:24–03:29). [investor-stated · strong] This is firm-wide history and retrospective thesis compression; later panel technical claims belong to the specialists. [researcher inference · strong]

16. Worth/Milken — AI market impacts (2024)

Publication and access. Worth Media uploaded the 4:02 Milken Global Conference interview on 2024-05-20. The video has provider automatic captions. The host asks; Mhatre answers. See source note.

Mhatre describes Lightspeed as technology-first, spanning formative and growth stages, and says ChatGPT changed public perception of computers as thought partners (00:26–02:25). [investor-stated · strong] At 03:20–03:53, he anticipates a taper after peak capital and uncertainty over which stack layers capture value. [investor-stated · moderate] This short conference interview offers a forecast, not a demonstrated allocation rule. [researcher inference · strong]

17. Bloomberg — AI moves from demos to deployment (2026)

Publication and access. Bloomberg uploaded the 10:21 interview on 2026-02-25; YouTube has manual English and automatic captions, while the DailyMotion page is a mirror. Mhatre is the guest; host premises remain host claims. See source note.

Mhatre predicts 2026 will move agents from demos to large deployments and emphasizes trust/reliability infrastructure (01:31–04:17). [investor-stated · moderate] He repeats promotional Anthropic revenue and safety-leadership claims; both need filings or independent verification. [investor-stated · provisional] The host's “165 AI companies” premise is firm-wide and not Mhatre's personal denominator. [researcher inference · strong]

18. CNBC International — software reborn (2026)

Publication and access. CNBC International uploaded the 6:54 interview on 2026-02-18. The video has provider automatic captions. Hosts question Mhatre about software repricing and Anthropic. See source note.

Mhatre argues agents commoditize some workflow construction while proprietary context/data remain costly, shifting software economics away from seat pricing (00:00–01:17). [investor-stated · strong] He pairs frontier intelligence with trust/safety as the Anthropic thesis (02:19–03:55). [investor-stated · moderate] This is an interested investor's current thesis, not comparative performance evidence. [researcher inference · strong]

19. CNBC — AI valuations and Anthropic (2025)

Publication and access. CNBC uploaded the 6:35 interview on 2025-03-14. The video has manual and automatic English tracks. Multiple anchors question Mhatre. See source note.

Mhatre says high AI prices reflect potential plus investor competition, then offers a training-cost versus inference-consumption margin thesis (00:45–03:29). [investor-stated · moderate] He expects multiple foundation-model winners, analogizing to hyperscalers (04:04–05:26). [investor-stated · provisional] These are forecasts from a fund with relevant positions. [researcher inference · strong]

20. Lightwork — thinking in future tense (2026)

Publication and access. Lightspeed published the 16:09 direct interview on 2026-06-17. The video has provider automatic captions. A firm host questions Mhatre; the setting is institutional marketing. See source note.

Mhatre says AI founders must design for a platform whose capability itself changes rapidly and demonstrate a continuously updating “future tense” value point (01:09–03:42). [investor-stated · strong] This updates the older static-inflection model: the wedge must survive a moving frontier. The model risks rewarding grand narratives unless paired with current customer proof. [researcher inference · strong]

21. DualEntry — the future of ERP (2026)

Publication and access. DualEntry uploaded the 1:44 multi-speaker investment video on 2026-02-05. The video has manual and automatic English tracks. Mhatre appears with company/investor speakers; caption labels are absent, so only clearly attributable segments are used. See source note.

Mhatre says many teams lacked ERP-specific, hands-on AI depth and frames cloud-to-AI as the next architecture shift (00:09–00:58). [investor-stated · moderate] Product benefit claims later in the montage are company/multi-speaker marketing, not reliably Mhatre speech. [contemporaneous record · provisional]

22. Cheddar — Rubrik IPO (2024)

Publication and access. Cheddar interviewed Mhatre for 4:51 on Rubrik's IPO day, 2024-04-25. The video has full provider captions; the host and Mhatre alternate clearly. See source note.

Mhatre credits continued product broadening around cloud data security and presents recoverability as a last line of defense (01:14–02:57). [investor-stated · strong] He says public-company stewardship matters to Global 2000 customers (04:05–04:40). [investor-stated · moderate] IPO-day enthusiasm and financial interest require filing-based controls; opening price is not realized return. [researcher inference · strong]

23. BCG — adapting to generative AI (2024)

Publication and access. Boston Consulting Group uploaded the 2:48 interview on 2024-05-10. The video has a short manual English track plus full automatic captions. A BCG host questions Mhatre. See source note.

Mhatre describes a horizontal/vertical opportunity map and says core selection ingredients remain visionary, determined people aimed at large markets where Lightspeed has enough understanding to lead early (00:49–02:39). [investor-stated · strong] The short format supplies selection language, not diligence mechanics or outcome evidence. [researcher inference · strong]

24. Film Session — student-athlete lessons for venture (2026)

Publication and access. Fisher Anderson uploaded the 4:00 interview excerpt on 2026-08-13. The video has provider automatic captions. Anderson asks; Mhatre answers. See source note.

Mhatre maps long-game belief, discipline and adversity tolerance from athletics to venture, then warns that team patterns are not mechanically portable across contexts (00:16–02:32). [investor-stated · strong] He analogizes startup accountability and differentiated roles to football (02:34–03:43). [investor-stated · moderate] This is philosophy, not evidence athletes outperform as founders or investors. [researcher inference · strong]

25. Geektime — identifying a billion-dollar company at idea stage (2015)

Publication and access. Geektime published a Hebrew edited Q&A on 2015-11-30. The full text is machine/reader translated for this analysis; it is not an English transcript and paragraph locators are used. See source note.

Mhatre contrasts lean 0→1 experimentation with rapid 1→∞ scaling after proof and describes a reported $1 million ThoughtSpot start (0→1 / ThoughtSpot answer paragraphs). [investor-stated · moderate] He says the firm invests in roughly one of 100 opportunities and wants to study misses (closing miss/funnel paragraphs). [investor-stated · provisional] Translation, selected interview editing and lack of a complete opportunity log prevent treating that as a verified denominator. [researcher inference · strong]

26. Bloomberg — Nutanix IPO reception (2016)

Publication and access. Bloomberg's official 2016-10-07 page identifies Mhatre, host Emily Chang and the topic, but no transcript/caption text is exposed through the page or public search routes. The embedded player did not yield a legitimate standalone audio route during this pass, so no transcription was attempted and no speech is reconstructed. See barrier note.

Only metadata is usable: a contemporaneous discussion of Nutanix and tech-IPO demand. [contemporaneous record · moderate] The page cannot support any quote, decision rationale, return figure or speaker proposition beyond its description. [researcher inference · strong]

27. Inc42 — the journey from founder to CEO (2018)

Publication and access. Dev Khare published the edited moderator notes on 2018-10-15, coauthored with Abhivyakti. The full article is not a verbatim transcript; Khare's synthesis and attributed Mhatre excerpts are separated. See source note.

The readout identifies adaptive learning, motivation, vision articulation and talent magnetism, then emphasizes phase-specific reinvention and hard personnel conversations (“Constant need…” and “Founder Traits…”). [retrospective witness · strong] The playbook says success patterns cluster more tightly than failure patterns; this is a heuristic, not a statistical demonstration. [researcher inference · strong]

28. Lightspeed/Rubrik event — setting crazy goals (2018)

Publication and access. Lightspeed's 2018-07-11 event page provides event context, images and speaker identities but little transcript-like substance. It credits Arif Janmohamed as page author and identifies Mhatre, John W. Thompson, Bipul Sinha, Janmohamed and Meredith Kendall. See source note.

The defensible evidence is documented convening around a portfolio company and operating mentor, not a direct Mhatre doctrine or quote. [documented behavior · strong] Event marketing cannot prove Mhatre's individual deal leadership, Thompson's precise contribution or Rubrik outcomes. [researcher inference · strong]

29. Axios View From the Top — enterprise AI ROI (2024)

Publication and access. Axios reported on 2024-12-19 from a December 16 sponsored event. The article includes a short direct-attributed Mhatre readout, distinct from the main Arvind Jain interview. See source note.

Mhatre says businesses would move from 2024 experimentation to evaluating applications and ROI in 2025. [investor-stated · strong] The single-sentence readout creates a falsifiable adoption milestone, but sponsorship and lack of full context limit depth. [researcher inference · strong]

30. Jefferies — venture capital's role in the age of AI (2025)

Publication and access. Jefferies' editorial team published the edited interview on 2025-06-02 from its April Private Growth Conference and explicitly says it was edited for length/clarity. See source note.

Mhatre frames AI as an expertise amplifier, argues deep-tech scale pushes venture toward an institutional capital-markets role, and tells companies to “buy forward” in product evolution (“Explosive Enterprise Growth,” “Role of Venture Capital,” and “Raising Capital”). [investor-stated · strong] The source demonstrates current doctrine, not the editorial introduction's unsupported Snap attribution or any audited return. [researcher inference · strong]

31. ETtech — AI bubble and India opportunity (2025)

Publication and access. ETtech published the edited joint interview on 2025-11-19. Samidha Sharma questions Mhatre and Bejul Somaia; answers are labeled and preserved by speaker. See source note.

Mhatre calls AI both a real once-in-a-generation shift and a valuation bubble, arguing durable technologies will solve real problems (valuation/bubble answers). [investor-stated · strong] He points to India's usage/talent density and sovereign/local-language motivations. [investor-stated · moderate] Portfolio-health, India-fund and application-layer answers belong largely to Somaia and cannot be converted into Mhatre claims; LSVP/LSIP remain distinct. [researcher inference · strong]

Asset and mirror reconciliation

The 25 substantive YouTube assets are 1uwzycO_E8c, U8VLDyrj6Mk, 3V_y7Y7xx9U, uehI2YhYc0g, psBR-vy8Vzk, SsGbBTBff9A, SsEoFDc1DIg, _AO2_Uzpdz4, N2mUVXkw41w, TXeGOCISrVg, X90SGz1BivU, S79P83FMZ0U, BF9UGJHjP7k, F7q5U4gQBV8, sBmP-MjSS2s, FAVz5IjyWks, oKZrrlK4oew, bDsVAE9cwmk, i3pdnKIoUC0, QocqJDTjGvA, M2NHdPYA8z0, zcgMBdQMC-8, fCmLZQ7FJ-A, t-h4NiKH7PM and cjeFOc5MolI. The two IIT assets are complementary edits from one appearance, so 25 assets become 24 families. [documented behavior · strong]

Cross-corpus synthesis

Seven recurring mechanisms survive source comparison:

1. Technical inflection plus exact wedge. Nutanix combines full-stack replatforming with simplicity (2015, 02:12–03:36); blockchain combines distributed architecture with digital money as initial pull (2024, 02:17–03:29). [investor-stated · strong]
2. Long vision, narrow entry. Geektime's lean 0→1 precedes 1→∞ acceleration (2015 Q&A); BCG still asks for a large market and enough domain understanding to lead (2024, 01:57–02:39). [investor-stated · strong]
3. Phase-specific company building. AppDynamics separates product fit, go-to-market and platform expansion (2018, 01:40–03:35); Inc42 adds founder reinvention and hard talent decisions (2018 notes). [retrospective witness · strong]
4. Adaptive learning over fixed prediction. Stanford centers listener/learner behavior (2020, 05:46); Newcomer says the operating math must recompile when gravity changes (2023, 02:51). [investor-stated · strong]
5. Moving-frontier underwriting. By 2026 “future tense” replaces a one-time platform shift: founders must update where value sits six and twelve months ahead (Lightwork, 02:20–03:42). [investor-stated · strong]
6. Trust as production infrastructure. NYSE frames security/misinformation risk (2024, 02:30–04:02); 2026 CNBC frames enterprise trust as a deployment moat (02:19–03:55). [investor-stated · strong]
7. Firm platform and team. StrictlyVC describes sequenced platform expansion (2017, 01:47); Hill & Valley and Rubrik-event sources show convening. [documented behavior · strong] Platform evidence does not assign every deal or outcome to Mhatre.

Dated chronology and evolution

• 2012: family science/operating exposure and SGI/Netscape company formation explain the technical-company lens (TiEcon). [investor-stated · strong]
• 2015: Nutanix, India and Geektime show architecture, ecosystem timing and lean 0→1; all three reduce later AI-era hindsight. [researcher inference · strong]
• 2017–2018: firm formation, venture-capital crowding, AppDynamics/MuleSoft phase transitions and founder adaptation become central. The 2017 capital-compression warning is an important pre-megafund baseline (10:54). [researcher inference · strong]
• 2020–2021: relationship/learning language broadens beyond deals; crypto adds architectural platform and balanced-regulation claims but also creates a later adverse test. [researcher inference · strong]
• 2023: post-SVB “gravity” makes regime change explicit and rejects plan invariance (Newcomer). [investor-stated · strong]
• 2024: AI moves from public realization toward trust, ROI and policy; blockchain is retrospectively recast as a technical platform. [researcher inference · strong]
• 2025: Mhatre simultaneously defends transformation and acknowledges a valuation bubble (ETtech). [investor-stated · strong]
• 2026: the doctrine becomes moving-frontier deployment: agents, outcome/context moats, trust infrastructure and “future tense.” These remain current forecasts, not mature outcomes. [researcher inference · strong]

Decision playbook

1. Name the technical discontinuity and the incumbent retrofit cost. Reject a category label without architecture evidence. Failure condition: incumbents match cheaply. [researcher inference · strong]
2. Find the smallest legible wedge. Keep 0→1 lean; demand a customer behavior that distinguishes broad need from friendly validation. Failure condition: pilots do not repeat. [researcher inference · strong]
3. Assess adaptive founder behavior, not only charisma. Look for changed beliefs, talent upgrades and hard conversations. Failure condition: the founder preserves identity at the expense of the phase. [researcher inference · strong]
4. Underwrite the next platform state. In AI, ask where capability/value move at six and twelve months while verifying today's use. Failure condition: the wedge is erased by model progress. [researcher inference · strong]
5. Recompile for gravity. Re-test burn, financing, buyer budgets and reserve needs after regime change. Failure condition: the plan depends on vanished capital. [researcher inference · strong]
6. Separate capability from production trust. Measure reliability, security and accountable deployment. Failure condition: demos do not become mission-critical workflows. [researcher inference · strong]
7. Attribute the platform honestly. Record named lead, partner, vehicle, adviser and founder contributions separately. Failure condition: firm marketing is mistaken for personal alpha. [researcher inference · strong]

Dated contradictions and belief updates

• Four founders versus six founding GPs. Mhatre's “all four of us” memory (2017) describes a core Stanford-classmate team; the 2000 WPG/Lightspeed account lists six founding GPs. Preserve scope/date instead of smoothing the conflict. [researcher inference · strong]
• Capital discipline versus platform scale. In 2017 Mhatre says more venture capital compresses returns (10:54); in 2025 he argues venture must become a larger institutional force (Jefferies). The update may reflect capital-intensive AI, but no evidence here resolves whether larger funds preserve returns. [researcher inference · strong]
• Crypto architecture versus cycle outcomes. The 2021 bullish forecast (03:39) and 2024 platform retrospective (02:17) require FTX-era and loss-denominator counterevidence; no Mhatre-led FTX claim is made. [researcher inference · strong]
• Independent-company aspiration versus AppDynamics sale. Company-building sources prize public readiness, while AppDynamics accepted Cisco's offer before trading. This is strategic optionality, not proof that independence was the only goal. [researcher inference · strong]
• AI hype versus AI transformation. By 2025 Mhatre explicitly says both can be true (ETtech); the discipline must therefore be company-specific proof, not denial of the bubble. [researcher inference · strong]
• Network access versus open selection. ThoughtSpot/repeat-founder access accelerates conviction, but the corpus offers no counterfactual for equally capable founders outside Stanford/diaspora/portfolio networks. [researcher inference · strong]

Five appearances to use first

1. Stanford 2020 — fullest official transcript and deepest process/biography; also the best demonstration that direct speech needs factual checking. [researcher inference · strong]
2. Newcomer 2023 — strongest regime-change language and post-2022 reflection, with full audio/captions. [researcher inference · strong]
3. Geektime 2015 — unusually specific 0→1/1→∞, ThoughtSpot and claimed funnel evidence; use with translation/denominator caveats. [researcher inference · strong]
4. StrictlyVC 2017 — contemporaneous firm strategy, AppDynamics formation and capital-crowding tension with multiple partner voices. [researcher inference · strong]
5. AppDynamics 2018 — clearest phase model and founder adaptation, recovered through explicitly documented OpenAI transcription after caption failure. [researcher inference · strong]

Unresolved record and access gaps

Bloomberg's 2016 Nutanix video remains metadata-only. No public LP letters, IC recordings, complete opportunity/pass archive, board-session archive or loss interview was found. Provider automatic captions contain name errors; three machine transcripts are research artifacts rather than official text. Edited interviews can omit qualifications; moderator/panel appearances often reveal agenda and network more than Mhatre propositions. The corpus is bounded public-web research, not an exhaustive lifetime appearance count, and it cannot establish personal returns, full portfolio causality or a complete opportunity denominator. [researcher inference · strong]

Model 1 — Detect the technical inflection and name the exact wedge

• Rule — Invest only when a dated technical change makes a previously impractical workflow newly possible and the company has an urgent, narrow entry product; “technology is growing” and a broad TAM are not sufficient.
• Evidence status — [investor-stated · strong] for technology-as-language and inflection reasoning; [documented behavior · moderate] for repeated application because original memos are unavailable and most case rationale is retrospective.
• Provenance and origin — Mhatre's Stanford talk says technical fluency lets an investor converse across domains and translate what becomes possible; his 2012 Zscaler quote ties cloud, mobility and security to a specific delivery change. Stanford transcript, pp. 1–2 · Zscaler round release
• Mechanism and reasoning — An inflection lowers cost, latency or complexity enough to change buyer behavior; the wedge creates a short feedback loop and distribution foothold. Without both, the investor is funding science or a slogan rather than a company. [researcher inference · strong]
• Historical and vehicle context — Enterprise/cloud waves from 2007–2014 offered architecture layers that long-duration Lightspeed vehicles could fund through several rounds. Current AI/chip/biology bets can require far more capital and faster re-underwriting; the same rule cannot imply the same check or reserve. Historical context [researcher inference · strong]
• When to use — Infrastructure, data, security, developer tools, applied AI, robotics and deep tech where a measurable technical constraint maps to a buyer/workflow. Do not use it when regulation, distribution or consumer behavior—not technical feasibility—is the dominant unknown.
• How to apply — Write the “before/after” architecture; name the constraint, buyer and first use; measure why now; identify incumbent/bundle and wait alternatives; set product/customer proof; kill if the technical delta does not create urgent willingness to switch/pay.
• Examples —
  • Nutanix — documented/retrospective application: simplified compute/storage architecture and an enterprise deployment wedge; chronology remains 2010–2011 and the original memo is absent. Nutanix lab
  • Zscaler — contemporaneous category evidence: cloud delivery and mobility supported a security inflection, but a Series C participation does not prove company formation credit. 2012 release
  • Materialize — interested-source application: streaming SQL without operational compromise names an architecture and developer wedge; private outcome and lead/economics remain unknown. Materialize note
  • Ricursive — in-scope counter/boundary: the 2026 chip-design loop names a technical inflection and wedge, but tape-out, transfer, buyer urgency and economics remain pre-proof; a real architecture thesis is not yet a validated investment edge. Ricursive note
• Evidence — Independent/adjacent controls include the Nutanix prospectus, the issuer's Zscaler round announcement, the complete Stanford transcript, and the 91-card ledger.
• Limits, failure modes, and exception ledger — Technical fluency can create false confidence; incumbents can bundle; a real inflection may accrue to another layer; customer switching may dominate architecture; outcome-aware essays select winners. The exception ledger must record later-stage entry (Zscaler), shared/team thesis (Materialize) and non-technical/consumer cases where this model should not be imposed.
• Evolution and contradictions — Enterprise-cloud examples made the rule visible in the 2010s; 2025–2026 writing extends it to frontier models, inference, workflows and chips. The Stanford science errors show fluency is not factual authority in every domain; specialists and external controls remain mandatory. Stanford caveat [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — AI makes prototypes and impressive demos cheap. Ricursive and Unconventional AI are grounded current inflection theses but pre-proof: require production/tape-out evidence, a defensible loop, independently verified workflow value, cost trajectory, safety/regulatory ownership and a wedge that survives model improvement. Ricursive note · Unconventional AI note [researcher inference · strong]

Model 2 — Hold the long vision; enter through a narrow door

• Rule — Demand a consequential long-term company vision, but fund and govern a narrow first wedge with explicit evidence gates before broad platform expansion.
• Evidence status — [investor-stated · strong] for extreme ambition/long vision; [researcher inference · strong] for the narrow-entry safety rule synthesized across cases and adverse controls.
• Provenance and origin — The 2016 Nutanix essay uses the “rocket ship to the moon” metaphor for purpose-built ambition; AppDynamics writing and founder testimony separate product, GTM and platform phases rather than treating the endpoint as present at formation. Nutanix essay · AppDynamics essay
• Mechanism and reasoning — Vision attracts founders, recruits and patient capital; a wedge creates urgency, learning and distribution. Expanding only after proof protects the company from a platform story that outruns product quality, GTM or capital. [researcher inference · strong]
• Historical and vehicle context — Multi-vintage venture/Select vehicles can support a long path and follow-ons, but their private duration/reserve terms are unknown. Cheap 2010s capital made broad expansion easier; current capital intensity and higher uncertainty demand explicit financing gates. Vehicle economics
• When to use — Early technical/enterprise products with a credible beachhead and multiple adjacent control points. Do not use “vision” to waive proof in roll-ups, consumer acquisition models, crowded feature markets or capital structures that cannot survive to the next gate.
• How to apply — Write the ten-year change, first buyer/workflow, minimum defensible product, reference-customer target and adjacent expansion map; prewrite what must be true before each expansion; reserve only against passed gates; kill if breadth degrades the wedge.
• Examples —
  • Nutanix — documented/retrospective application: large architecture ambition paired with enterprise deployment proof; later IPO does not validate every original premise. Nutanix lab
  • AppDynamics — founder-witness application: monitoring wedge preceded GTM and platform phases; the strategic sale before IPO complicates the ultimate public-company narrative. AppDynamics lab
  • Rubrik — documented behavior: data-management wedge expanded through repeated firm backing; the 2026 proxy confirms vehicles/board, not which milestone drove each follow-on. Rubrik lab
  • BYJU'S — countercase: global-platform growth without a surviving cohort/cash/governance gate record shows why ambition cannot be the scorecard. Failure matrix
• Evidence — The direct Nutanix essay, founder/investor AppDynamics corpus, issuer AppDynamics S-1, and Rubrik proxy independently anchor phases, outcomes and attribution.
• Limits, failure modes, and exception ledger — Vision is cheap after success; a narrow wedge can cap the market; platform expansion can be necessary before conventional metrics; incumbents can absorb the adjacency; a well-funded story can hide weak cohorts. Exceptions must state when regulation, deep-tech milestones or network formation require a different first proof.
• Evolution and contradictions — Early essays celebrated ambitious enterprise builders and public-company endpoints. AppDynamics' sale, BYJU'S adverse outcome and current AI capital needs add explicit governance, financing and stop gates that are not prominent in the original rhetoric. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — Temporal and Fireworks are grounded current wedge/platform claims, not outcomes. Write separate scorecards for production task success, workflow switching, customer ROI and reliability before breadth; require a financing plan that survives inference-cost and capability shifts. Temporal note · Fireworks note [researcher inference · strong]

Model 3 — Run lean from zero to one; accelerate only after proof

• Rule — Before product-market proof, minimize irreversible hiring, spend and complexity; after a named proof gate, accelerate only the functions the next phase requires.
• Evidence status — [investor-stated · moderate] because the clearest formulation is an edited/translated 2015 interview; [retrospective witness · strong] for AppDynamics' phase sequence; actual company budgets and Mhatre decisions remain unavailable.
• Provenance and origin — In Geektime, Mhatre describes a lean zero-to-one phase followed by acceleration; Bansal later describes product-market fit, go-to-market and platform as distinct company-building phases. Geektime note · AppDynamics note
• Mechanism and reasoning — Lean design extends runway and preserves options while the company learns; a proof gate concentrates capital around validated pull. “Lean” is not low ambition—it is matching organizational complexity and cash burn to evidence. [researcher inference · strong]
• Historical and vehicle context — Seed software companies in the 2010s could often reach product/customer proof cheaply; frontier AI, chips, robotics and biology can require capital before revenue. Long-duration Lightspeed vehicles may absorb that difference, but reserve and fund terms are private. Vehicle context
• When to use — Software/developer products with short build/test loops and a measurable adoption milestone. Adapt—not mechanically apply—to regulated/deep-tech bets where prototype, clinical, fabrication or safety milestones require substantial pre-revenue capital.
• How to apply — Define zero-to-one proof; fund runway to it with contingency; assign one owner per unknown; forbid premature platform/geography expansion; at gate, choose stop, repeat or accelerate; write the new org/hiring/GTM plan and reserve before spending.
• Examples —
  • AppDynamics — founder-witness application: Bansal separates PMF, GTM and platform; no public record proves Mhatre set the budgets. AppDynamics note
  • ThoughtSpot — reconstructed application: 2012 Seed and 2014 Series A are distinct events; product/search proof should precede enterprise scaling, but terms/milestones are unavailable. ThoughtSpot lab
  • Rubrik — follow-on illustration: repeated vehicles imply acceleration after progress, while the precise gates and cost chain remain private. Rubrik proxy
  • BYJU'S — countercase: later global expansion/fundraising without a public cohort/cash/governance gate record shows how acceleration can become its own evidence. Failure matrix
• Evidence — The edited Geektime source, founder/investor AppDynamics audio evidence, issuer ThoughtSpot 2014 release, and regulator/court-controlled BYJU'S lab provide independent phase and countercase anchors.
• Limits, failure modes, and exception ledger — Lean can starve security, quality or regulatory work; a competitor can capture the market while a company waits; “proof” can be vanity usage; capital-heavy systems need technical milestones instead of SaaS revenue. The exception ledger must state the capital model, irreversibility, required pre-proof spend and why conventional lean gates do not fit.
• Evolution and contradictions — The 2015 idea-stage rule was formulated for lean software formation. Current AI/deep-tech cards include late/private rounds and capital-intensive science; the transferable part is staged irreversible commitment, not a universal small check. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — Unconventional AI is a grounded capital-intensive current boundary, not a validated application. Define proof as repeated technical/production milestones and paid workflow change—not demo quality—and separate experimental compute from scalable unit economics. Unconventional AI note [researcher inference · strong]

Model 4 — Back adaptive founders and build the recruiting magnet

• Rule — Select founders for purpose plus demonstrated learning; then judge whether they can recruit the complementary people required at product, GTM, platform and governance phases.
• Evidence status — [investor-stated · strong] for founder purpose/adaptability; [documented behavior · moderate] for repeated application because founder stories are selected and recruiting attribution is mostly private.
• Provenance and origin — Mhatre's Stanford and company-building accounts emphasize founders who learn and pursue extreme goals; the AppDynamics corpus explicitly separates the skills required as the company evolves. Stanford transcript, pp. 1–3 · AppDynamics essay
• Mechanism and reasoning — A technical founder provides insight and product velocity; learning behavior prevents the original insight from becoming dogma; recruiting expands the company's capability faster than one person can. The board adds value only if specific gaps and evidence are visible. [researcher inference · strong]
• Historical and vehicle context — Elite firm brands, repeat-founder networks and multi-partner teams create recruiting/access advantages not available to every vehicle. This model must separate founder quality from the platform that helped attract talent. People map
• When to use — Early company formation and phase transitions where leadership behavior is observable. At later stage, replace charisma tests with operating evidence, executive-team quality, governance information and reference checks.
• How to apply — Ask for a changed-mind example; reference good and bad periods; map current/missing capabilities; test bad-news transparency; prewrite phase-specific hires; assign board/recruiting owner; kill if the founder blocks information, cannot recruit or treats adaptation as betrayal.
• Examples —
  • AppDynamics — founder-witness application: Bansal describes different PMF/GTM/platform phases; public record supports Mhatre's board date, not which recruit or intervention he owned. AppDynamics lab
  • MuleSoft — founder relationship: Mason credits Mhatre with initial/subsequent backing and the filing records board service; the recruitment/board action log is private. Ross Mason account
  • ThoughtSpot — repeat-founder test: network familiarity and firm-led financing can improve access but heighten confirmation bias; personal lead and ownership remain unknown. ThoughtSpot lab
  • Glean familiarity — researcher-applied countercase: repeat enterprise-search relationships improve people information but can turn friendly-customer signals into false PMF or suppress an outside view; the public record lacks the contemporaneous reference and dissent log needed to show the control worked. Glean note
• Evidence — Direct support comes from Stanford and Mhatre's AppDynamics essay; independent role controls come from the AppDynamics S-1 and MuleSoft prospectus.
• Limits, failure modes, and exception ledger — Adaptability can become post-hoc praise; purpose can become identity defense; repeat networks bias selection; executive replacement can destroy trust; the investor may claim founder work done by others. Log founder evidence, references, missing capability, recruiter/board owner, intervention and outcome by date.
• Evolution and contradictions — Early rhetoric emphasized purpose and extreme goals; later gravity/feedback language makes adaptation explicit. BYJU'S and AI safety/governance show that adaptability must include transparency and institutional controls, not just product/GTM learning. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — Ineffable is a grounded current people bet but not an outcome. Differentiate AI founders through technical judgment, customer learning, safety/governance behavior, capital discipline and recruiting across models, infrastructure, product and domain expertise. Move 37 note [researcher inference · strong]

Model 5 — Replace static prediction with explicit feedback loops

• Rule — Prewrite evidence that can change thesis, capital or stop decisions; distinguish the investor's update loop from a company's self-reinforcing product/developer/data loop and prove each separately.
• Evidence status — [investor-stated · strong] for feedback over prediction; [researcher inference · moderate] for company-flywheel application because public unit economics and causal tests are thin.
• Provenance and origin — Mhatre's 2023 Newcomer appearance contrasts prediction with feedback under changing “gravity”; Snorkel, Alchemy and Ricursive writings describe learning/developer loops, but they are firm/company narratives. Newcomer, 07:45–11:10 · Ricursive note
• Mechanism and reasoning — An investor loop updates beliefs from milestones. A company flywheel compounds value when more usage improves product/data/ecosystem and that improvement creates more usage at attractive economics. Conflating them turns “we learned” into an untested moat claim. [researcher inference · strong]
• Historical and vehicle context — Long-duration follow-on vehicles can fund repeated loops; this creates sunk-cost risk. Crypto cycles, chip tape-outs and model changes can interrupt apparent compounding. Vehicle record
• When to use — High-uncertainty categories with measurable iterations. Do not call a loop when growth comes from paid acquisition, capital subsidy, market beta or one-time data ingestion.
• How to apply — Write hypothesis, next observation, update threshold and decision owner; for a company flywheel map input→improvement→user value→retention/revenue→new input; measure lag/cost; kill if iteration does not improve independent customer or economic evidence.
• Examples —
  • AppDynamics — investor/company update illustration: product, GTM and platform are separate evidence stages, not one success loop. AppDynamics note
  • Snorkel AI — company-loop hypothesis: programmatic labeling/data development can improve iteration, but founder technical claims dominate the session and economics remain private. Snorkel note
  • Ricursive — provisional flywheel: AI for chip design and chips for AI is an explicit loop thesis; tape-out, transfer and economic evidence are immature. Ricursive note
  • BYJU'S — countercase: funding/growth without a public cohort/cash update record can masquerade as evidence. Failure matrix
• Evidence — The Newcomer direct appearance, founder-heavy Snorkel session, interested Ricursive writing, and regulator/court-controlled BYJU'S record supply distinct evidence families.
• Limits, failure modes, and exception ledger — Metrics can be selected; loops can be negative; learning may not compound economically; follow-ons can reward sunk cost; outcome-aware accounts can invent gates. Log predicted metric, observation date, independent source, decision change, company-loop unit economics and alternative explanation.
• Evolution and contradictions — Earlier prediction essays listed trends; later gravity language explicitly demotes static forecasting. Crypto forecasts and recent AI claims show the update: category belief must not substitute for company-loop proof. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — Ricursive is a grounded current loop hypothesis, not validation. Require production telemetry and outside customer references; test tape-out/transfer, data rights, model substitution, inference cost and whether usage improves a proprietary system rather than a third-party model. Ricursive note [researcher inference · strong]

Model 6 — Re-underwrite when the gravity regime changes

• Rule — When capital cost, buyer demand, technical capability, regulation or exit conditions change, rebuild budgets, proof gates, financing and valuation from first principles; do not merely extend runway or repeat old tactics.
• Evidence status — [investor-stated · strong] for the gravity metaphor and cycle recollection; [researcher inference · provisional] for consistent portfolio application because budget/IC records are private.
• Provenance and origin — In Newcomer, Mhatre frames dot-com survival as adapting to changed gravity; the 2017 cofounder interview recounts first-fund/cycle experience. Newcomer, 02:01–11:10 · StrictlyVC/TechCrunch
• Mechanism and reasoning — A strategy calibrated to cheap capital or expanding budgets fails when financing/buyer conditions change. Recompilation makes the change operational: cash plan, milestone, ownership, reserve and stop decision—not a retrospective metaphor. [researcher inference · strong]
• Historical and vehicle context — Lightspeed/WPG formation crossed dot-com; MuleSoft/AppDynamics crossed GFC; Rubrik crossed 2020–2022; AI bets face large rounds and fast technical resets. Vehicle duration/reserves determine available responses. Historical context
• When to use — Trigger on material financing, demand, capability, regulatory or liquidity discontinuities. Do not invoke it for ordinary variance or to rationalize an outcome after the fact.
• How to apply — Define trigger; reforecast runway and proof; rerun downside ownership/financing; rank keep/cut/accelerate; assign board/IC decision; record why the new plan differs; set date for next recompilation.
• Examples —
  • Lightspeed formation — institutional application: WPG continuity/six GPs and the later four-core story crossed the dot-com reset; exact company tactics are not public. Formation record
  • AppDynamics — researcher-applied illustration: public readiness ended in strategic sale, a liquidity change that illustrates—but does not document—Mhatre's gravity re-underwriting. AppDynamics lab
  • Rubrik — researcher-applied illustration: 2014–2024 company building crossed cheap capital and reset; holdings are known, but no public budget or IC record proves gravity re-underwriting. Rubrik proxy
  • BYJU'S — countercase: public record does not show which warning changed capital/governance/stop decisions, so the model cannot be claimed as applied. Failure matrix
• Evidence — Direct cycle evidence comes from Newcomer and TechCrunch/StrictlyVC; independent transaction/failure controls come from Cisco's AppDynamics close and the BYJU'S process.
• Limits, failure modes, and exception ledger — Gravity is flexible and can explain any pivot; macro calls can crowd out company evidence; cutting can destroy an inflection opportunity; large funds can postpone recognition. Log trigger, prior/new plan, date, owner, changed capital and falsifier.
• Evolution and contradictions — 2017 accounts emphasize surviving the first cycle; 2023 formalizes gravity/feedback; 2025–2026 AI appearances accept large capital while warning about deployment/value. No public record proves a standardized re-underwriting protocol. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — Current Jefferies AI remarks ground the changed regime but not a documented portfolio recompilation. Recompile when model capability, inference economics, regulation or capital availability crosses a prewritten threshold; require customer/workflow evidence to choose accelerate versus stop. Jefferies note [researcher inference · strong]

Model 7 — Use the firm platform; keep a case-level attribution ledger

• Rule — Treat partners, specialists, adviser entities, vehicles and founder networks as real investment capabilities, while assigning source, sponsor, diligence, approval, negotiation, board, recruiting, follow-on and exit credit by case.
• Evidence status — [documented behavior · strong] for multi-partner cards/filings/entities; [researcher inference · strong] for the attribution-ledger control; private function logs remain unavailable.
• Provenance and origin — WPG continuity and six founding GPs constrain the four-classmate narrative; later CEO council, health/crypto/growth writings and company cards show a multi-specialist platform. Light Reading · Authorship controls
• Mechanism and reasoning — Specialization expands sourcing/diligence/support; pooled vehicles provide duration/reserves; networks recruit. An attribution ledger prevents platform value from becoming star-partner mythology or legal-entity error. [researcher inference · strong]
• Historical and vehicle context — LSVP, LSIP and Faction are distinct adviser perimeters; numbered funds/Select/SPVs and shared GP power vary by case. Legal page · Vehicle record
• When to use — Every institutional/team investment and especially cross-border, multi-stage or multi-specialist cases. A solo investor should use the same function map to identify missing capability rather than imitate platform breadth.
• How to apply — Freeze adviser/vehicle; name owner for every function; record coauthors and founder witnesses; attach dated board/holder evidence; separate firm/personal economics; revisit the ledger at follow-on and exit.
• Examples —
  • MuleSoft — strong person attribution inside a platform: founder account says Mhatre led initial/subsequent; vehicles/shared power remain separate. MuleSoft lab
  • Rubrik — lead plus vehicle/personal separation: card says Mhatre led/board; proxy separates five vehicles and Mhatre lines. Rubrik lab
  • Netskope — counter to star credit: Lightspeed led and card calls Arif/Mhatre advisers; Arif—not Mhatre—was director. Netskope ledger row
  • BYJU'S — entity/adverse countercase: LSVP and LSIP both appear; no public source allocates Mhatre lead/board/loss. BYJU'S lab
• Evidence — Independent controls include Light Reading's formation report, Lightspeed legal disclosures, issuer MuleSoft prospectus, and Rubrik proxy.
• Limits, failure modes, and exception ledger — Team pages can be marketing; shared power does not show IC votes; platform breadth creates handoff/capacity/conflict risk; partner credit can follow outcomes. Log unresolved function, conflicting claim, entity, evidence date and economic boundary.
• Evolution and contradictions — Formation history moves from institutional continuity/six GPs to a four-core cofounder narrative; modern coauthored AI/health/crypto work makes distributed expertise visible but can still blur decision rights. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — Fireworks' five-author investment post grounds the multi-specialist pattern but not function ownership. Publish a responsibility map and capacity budget; measure founder service by named work rather than platform headcount. Fireworks note [researcher inference · strong]

Model 8 — Use repeat relationships for people information; force an outside-view check

• Rule — Treat a repeat-founder or incubated relationship as richer evidence about behavior, not as proof of market, price or governance; require an independent countercase and fresh diligence.
• Evidence status — [documented behavior · moderate] for repeat relationships; [researcher inference · strong] for the outside-view control because original relationship decisions are mostly retrospective/interested.
• Provenance and origin — ThoughtSpot and Rubrik writings foreground known founders/team formation; Glean writings repeat the enterprise-search network. ThoughtSpot essay · Rubrik essay
• Mechanism and reasoning — Repeated work reveals integrity, learning and recruiting behavior, reducing people uncertainty. It simultaneously creates familiarity, access and confirmation bias; fresh market/product/terms evidence must remain independent. [researcher inference · strong]
• Historical and vehicle context — A large firm with alumni, founders and multi-stage capital sees repeat opportunities a first-time fund may not. That access advantage is not selection or return proof. People map
• When to use — Repeat founders, operator spinouts and incubations. Do not use relationship quality to waive reference checks, conflicts, entity allocation, price or the changed market.
• How to apply — Separate known-person evidence from unknown-company evidence; obtain disconfirming references; re-map market/competition; assign a devil's advocate; compare terms with a cold opportunity; log which prior behavior predicts the new phase.
• Examples —
  • ThoughtSpot — researcher-applied familiarity counter-control: repeat-founder/access evidence may improve people information, but the card's 2012 Seed versus issuer's 2014 Series A and missing dissent log show why chronology and outside-view diligence cannot be waived. Lab
  • Rubrik — documented relationship/role evidence; outside-view application unknown: Bipul Sinha/Lightspeed formation and explicit Mhatre lead/board; firm narrative remains interested. Lab
  • Glean — researcher-applied illustration: repeated enterprise-search writings show network/thesis continuity; role/economics remain team/private. Glean note
• Evidence — Independent controls include the ThoughtSpot issuer release, Rubrik proxy, and the interested Glean and Rubrik writings; no complete outside-view or dissent log was found.
• Limits, failure modes, and exception ledger — People information decays across role/market; loyalty can block leadership change; access creates adverse selection; incubations blur firm/founder credit. Log prior relationship, observed behavior, fresh unknowns, independent reviewer, conflict/allocation and outcome.
• Evolution and contradictions — Early repeat enterprise relationships became a platform advantage; current multi-partner AI/deep-tech teams make “repeat” more institutional and less person-specific. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — AI labs and spinouts make pedigree/network abundant. Use Ineffable's people-bet framing only with fresh technical/governance evidence and outcome immaturity explicit. Move 37 note [researcher inference · strong]

Model 9 — Engineer independent readiness; keep IPO and sale optional

• Rule — Build governance, controls, product durability and unit economics for independent/public-company readiness; treat IPO, sale or continued private operation as later options rather than the initial thesis.
• Evidence status — [investor-stated · strong] for public-company durability; [researcher inference · moderate] for readiness as option generation because outcome comparisons do not reveal the actual exit deliberations.
• Provenance and origin — 2016 enterprise-cloud/IPO essays celebrate durable public companies; AppDynamics filed and then sold, while Nutanix and Rubrik listed. Writings chronology · AppDynamics close
• Mechanism and reasoning — Readiness increases strategic choice and financing credibility. Equating readiness with IPO confuses company quality, market window and investor liquidity. [researcher inference · strong]
• Historical and vehicle context — Exit windows changed across 2016–2026; multi-vintage funds may have different liquidity pressures. Issuer IPO proceeds and EV are not investor returns. Liquidity controls
• When to use — Scaled companies approaching material financing/exit choices. Do not force seed companies to optimize for public controls before proving a wedge.
• How to apply — Build audit/control, governance, retention/unit-economics and leadership checklist; model independent/downturn plan; compare IPO/sale/private options by stakeholder/security; log board decision and liquidity separately.
• Examples —
  • Nutanix — documented outcome; researcher-applied illustration: completed IPO; filing gives dated vehicle/board evidence, not realization. Lab
  • AppDynamics — documented outcome; researcher-applied illustration: filed but sold to Cisco; readiness did not mandate IPO. Lab
  • Rubrik — documented outcome; researcher-applied illustration: public with separate issuer/vehicle/personal lines. Lab
  • ThoughtSpot — documented status; researcher-applied counter/control: still Private on the frozen card; a financing valuation is not readiness, liquidity or return. Lab
• Evidence — Independent filings/transactions include Nutanix 424B4, AppDynamics S-1, Cisco close, and Rubrik proxy.
• Limits, failure modes, and exception ledger — Public readiness can create bureaucracy; sale can maximize or destroy long-term option; market windows and preference stacks differ; cards can be stale. Log readiness evidence, alternatives, stakeholder/security impact, board role and actual liquidity.
• Evolution and contradictions — 2016 essays foreground IPO/public durability; AppDynamics and later stale card statuses force a company-readiness rather than endpoint interpretation. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — For late AI companies such as Anthropic, private scale/valuation is not public readiness. Test audited revenue quality, compute commitments, safety/governance, customer concentration and financing resilience. Anthropic note [researcher inference · strong]

Model 10 — Treat trust, safety, reliability and recoverability as adoption infrastructure

• Rule — In systems controlling data, identity, security or agent workflows, safeguards are product infrastructure and adoption prerequisites, not post-scale compliance features.
• Evidence status — [investor-stated · strong] for safety/scalability and security trust; [researcher inference · strong] for the unified adoption-infrastructure rule.
• Provenance and origin — Zscaler/Rubrik theses center trustworthy enterprise control; Glean emphasizes permissions; Anthropic writing rejects a safety/scalability tradeoff; Temporal emphasizes recoverable workflows for fallible agents. AI safety note · Temporal note
• Mechanism and reasoning — Enterprise adoption requires controlled access, uptime, auditability and recovery. Safeguards reduce expected failure cost and make delegated automation usable; they can also impose latency/cost, so performance must be measured. [researcher inference · strong]
• Historical and vehicle context — Cloud security/data systems made trust central; AI agents increase blast radius and regulatory scrutiny. Specialist/platform diligence matters, but responsibility cannot be outsourced to a coauthor. [researcher inference · strong]
• When to use — Security, data, identity, health, fintech, frontier AI and autonomous agents. Do not use generic “safety” branding without threat model, owner, measurable controls and incident process.
• How to apply — Map assets/actors/failures; define reliability/SLO, permissions, audit, recovery and safety tests; run adversarial diligence; assign board owner; gate rollout/capital on independent evidence; stop on concealed incidents or unbounded failure.
• Examples —
  • Zscaler — documented thesis; researcher-applied illustration: cloud security adoption depends on uptime/threat efficacy and trust. Lab
  • Rubrik — documented outcome; researcher-applied illustration: data recovery/security and public-company controls are core; security failures are kill criteria. Lab
  • Temporal — investor-stated current thesis; outcome immature: durable workflows are positioned as control for fallible agents. Note
  • Anthropic — investor-stated in-domain counter/boundary: the safety/scalability thesis is direct, but public incident, retention and independently measured adoption evidence is too immature to validate the rule. AI safety note
• Evidence — The Zscaler issuer release, Rubrik proxy, direct AI safety writing note, and Temporal note are distinct source families; no in-domain public portfolio failure application was located.
• Limits, failure modes, and exception ledger — Safety claims can be marketing; controls can bundle into incumbents; compliance can slow iteration without reducing risk; no system is failure-free. Log threat model, test, evidence owner, residual risk, incident and adoption/economic effect.
• Evolution and contradictions — The rule evolves from security/recovery to permissions and AI-agent safety. Stanford scientific errors reinforce that governance needs domain experts and correction, not investor confidence alone. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — Anthropic/Temporal are current grounded theses, not validated outcomes: require deployment incidents, evals, permissions, rollback/recovery and independently measured customer trust before scaling. Anthropic note [researcher inference · strong]

Model 11 — Prefer proprietary context, permissions and workflow control over generic model capability

• Rule — In applied AI, underwrite control of authorized context, permissions and a recurring workflow before treating access to a capable third-party model as a moat.
• Evidence status — [investor-stated · strong] for context/workflow theses in Glean, Typeface and current AI appearances; [researcher inference · moderate] for durable moat because retention, gross margin and model-substitution tests are mostly private.
• Provenance and origin — The enterprise-search and generative-work essays move from discovering trusted knowledge to generating inside governed workflows; current interviews emphasize deployment value rather than model spectacle. Glean note · Typeface note · Jefferies note
• Mechanism and reasoning — Authorized enterprise context improves task relevance; permissions and workflow state make outputs usable; repeated integration creates switching cost. Generic capability alone is vulnerable to foundation-model improvement or incumbent bundling. [researcher inference · strong]
• Historical and vehicle context — Earlier enterprise bets controlled databases, integration or observability layers; foundation models move capability outside the application company and raise data-rights, inference-cost and platform-dependency risks. Writings evolution [researcher inference · strong]
• When to use — Enterprise search, copilots, agents and workflow software where customers own differentiated data/process. Do not use it for open-web consumer tasks or products whose only differentiation is a model API prompt.
• How to apply — Map data rights, permission propagation, workflow write-back and system-of-record ownership; test model substitution, customer retention and gross margin; kill if the product loses value when the base model or incumbent bundle improves.
• Examples —
  • Glean — investor-stated current thesis; application unproved: trusted enterprise knowledge plus permissions is the asserted control layer; public retention/economic evidence remains interested. Glean note
  • Typeface — investor-stated current thesis; application unproved: governed brand context and content workflow extend the thesis, with explicit co-writing/AI authorship boundaries. Typeface note
  • ThoughtSpot — researcher-applied illustration: enterprise search supplies structured data context, but current role/economics and durable moat evidence remain incomplete. ThoughtSpot lab
  • DualEntry — investor-stated current illustration; outcome unproved: accounting workflow supplies a narrow structured-context example; the 1:44 appearance provides no independent retention or margin proof. DualEntry note
  • Generic-model bundling — hypothetical countercase: if an incumbent supplies equivalent permissioned context and write-back, the application can become a feature; no complete churn/margin countercase is public. Glean and Typeface premise controls [researcher inference · moderate]
• Evidence — Independent issuer context comes from the ThoughtSpot 2014 release; direct/firm sources include Glean, Typeface, and the DualEntry appearance. These support the thesis, not a realized moat.
• Limits, failure modes, and exception ledger — Customers may not grant data rights; permission sync can fail; workflows can be shallow; incumbents can bundle; inference cost can erase margin; stronger models can reduce proprietary advantage. Log model/provider, authorized corpus, workflow depth, substitution test, retention, margin and security incident.
• Evolution and contradictions — Business-intelligence/search writing treated data access as the wedge; generative AI shifts emphasis to governed context, permissions and action. Current claims are newer and more interested than the mature enterprise outcomes. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — Glean and Typeface are grounded current examples, not proof: diligence one permissioned production workflow, swap the underlying model, measure value/retention and verify data lineage before paying an AI-platform multiple. Glean note · Typeface note [researcher inference · strong]

Model 12 — Underwrite geography through local behavior, talent, infrastructure and entity fit

• Rule — Treat geography as a bundle of local user behavior, distribution, talent, infrastructure, regulation and legal-entity allocation—not as a generic growth adjective or one partner's personal market.
• Evidence status — [contemporaneous record · strong] for the 2006 India-office plan and distinct adviser perimeters; [researcher inference · moderate] for a consistently applied geography model because sourcing/IC/allocation records are private.
• Provenance and origin — Contemporaneous reporting records Mhatre discussing an India office; ShareChat writing describes India-specific mobile behavior; current legal/card evidence separates LSVP and LSIP. Economic Times, 2006 · ShareChat note · Legal page
• Mechanism and reasoning — Local infrastructure and behavior create different wedges; local teams improve sourcing and support; correct adviser/entity allocation determines mandate, governance and economics. Global pattern recognition helps only after these local differences are made explicit. [researcher inference · strong]
• Historical and vehicle context — The 2006 office plan predates the current distinct-adviser legal perimeter. Udaan, Yellow.ai and BYJU'S cards expose dual LSVP/LSIP associations, so website unity cannot erase entity, year or team differences. Ledger
• When to use — Cross-border sourcing, local consumer/distribution models and globally sold technical products built from a different talent base. Do not use country narratives to waive product economics, governance or allocation conflicts.
• How to apply — Write the local constraint/behavior, customer and distribution; map talent/infrastructure/regulation; freeze adviser/vehicle and team; compare local and global competitors; set currency/financing/governance gates; retain a local adverse case.
• Examples —
  • ShareChat — investor-stated/coauthored local-thesis illustration: India-specific mobile-language behavior supports a local product thesis, but the firm/India round does not prove Ravi led it. ShareChat note
  • Udaan — documented entity illustration: its live card shows parallel LSVP/LSIP 2016 Series A associations and a multi-person team; function and allocation remain unknown. Ledger
  • Yellow.ai — documented entity illustration: its card shows different LSVP 2020 and LSIP 2019 Series A lines, making entity/year separation substantive rather than clerical. Ledger
  • Nutanix — researcher-applied cross-border illustration: globally sold infrastructure and an India-linked founder/talent story illustrate technical formation; the case is not an India consumer template. Nutanix lab
  • BYJU'S — documented adverse countercase: a global education story cannot replace cohort, cash, governance and adviser-specific attribution; no personal Mhatre loss/decision record is public. BYJU'S lab
• Evidence — Distinct source families include contemporaneous Economic Times, the coauthored ShareChat work, current Lightspeed legal disclosure, and regulator/court-controlled BYJU'S record.
• Limits, failure modes, and exception ledger — Country narratives stereotype users; global enterprise products may be location-neutral; currency/regulation and follow-on markets shift; adviser mandates overlap; team pages can blur credit. Log customer geography, build geography, adviser/vehicle, allocation rationale, local owner, currency/regulatory exposure and comparable cohort.
• Evolution and contradictions — A 2006 office-expansion quote becomes a later multi-adviser platform; direct writings on local behavior coexist with dual-entity cards. Exact India-platform formation and case allocation beyond disclosed pages remain unresolved. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — For cross-border AI such as Yellow.ai, test local language/data rights, inference infrastructure, buyer budget and entity allocation separately; the frozen card is current association evidence, not a validated outcome or personal lead. Yellow.ai ledger row [researcher inference · strong]

Model 13 — Separate company importance from security price, ownership, vehicle and return

• Rule — Score company quality and historical importance separately from entry price, security terms, ownership, dilution, vehicle allocation, realized cash flow and personal economics; never let an impressive company or valuation stand in for return.
• Evidence status — [documented behavior · strong] for the security/vehicle distinctions in filings; [researcher inference · strong] for the decision control; portfolio return application remains [researcher inference · insufficient public record] without full cash flows.
• Provenance and origin — Nutanix, MuleSoft, Rubrik and Netskope filings expose security holders and dated ownership, while transaction releases expose issuer proceeds or enterprise value. The package's central accounting control is to keep those quantities separate. Portfolio forensics · Rubrik proxy
• Mechanism and reasoning — Venture returns depend on price, security, ownership, dilution, reserves, allocation and timing of cash flows. A valuable company can be a poor investment at the wrong terms, and a large issuer event can yield different outcomes across vehicles and people. [researcher inference · strong]
• Historical and vehicle context — Numbered funds, Select funds, SPVs and separately disclosed Mhatre/family positions coexist. Filing snapshots are dated; voting/dispositive power is not economic ownership; issuer IPO proceeds and acquisition EV are not investor proceeds. Vehicle economics
• When to use — Every initial/follow-on/secondary/exit and every public ranking. Do not infer zero from a missing sub-threshold holder or compute personal return from shared vehicle power.
• How to apply — Freeze date, adviser, vehicle, security, check, price, shares, ownership and dilution; model reserves/downside; record sales/distributions/residual value by vehicle; calculate MOIC/IRR only from dated cash flows; retain unknowns rather than reverse-engineering headlines.
• Examples —
  • Nutanix — documented accounting application: the filing separates LVP VII/VIII holdings; Forbes reports a firm/vehicle paper mark, not realization or Mhatre economics. Nutanix lab
  • Rubrik — documented accounting application: the 2026 proxy separates five Lightspeed vehicles from Mhatre direct/family Class A holdings; neither supplies cost basis or realized return. Rubrik lab
  • Netskope — documented accounting application: a disclosed post-offer position and offer-price paper value are vehicle exposure, while Arif—not Ravi—held the board seat. Directional ranking
  • BYJU'S — documented adverse countercase: CIRP is a company/adviser adverse outcome, but stake, write-down, loss, Mhatre decision and cash flows are not public. Postmortem
• Evidence — Independent regulator/transaction controls include Nutanix 424B4, MuleSoft 424B4, Rubrik proxy and Netskope final prospectus.
• Limits, failure modes, and exception ledger — Private securities and side letters are hidden; filings omit sales and sub-threshold positions; paper marks move; shared control obscures allocation. Log missing basis, dilution, sale/distribution, residual value, carry and cross-vehicle allocation; label any ranking directional.
• Evolution and contradictions — Early winner narratives foreground company outcomes; later public filings make vehicle/personal separation possible. Recent AI valuations are the strongest temptation to collapse company promise into investment return. [researcher inference · strong]
• Modern VC translation (as of 2026-08-23) — Anthropic and other late private AI rounds are grounded current tests: require security-level price/ownership/reserve/downside by vehicle and do not treat headline valuation or firm participation as Mhatre return. Anthropic note [researcher inference · strong]

Model 14 — Choose deliberately between an enabling control layer and a full-stack operator

• Rule — Decide whether value should accrue to an enabling layer used across workflows or to a vertically integrated operator controlling the end outcome; do not mix the two capital, distribution and governance models.
• Evidence status — [researcher inference · moderate] as a cross-case synthesis; Mhatre's writings support both company types but no public source states a general allocation rule, and matched outcomes are incomplete.
• Provenance and origin — Yugabyte, Glean and Fireworks describe enabling infrastructure/control layers; Blend, Multiverse and Dexterity describe operators that absorb workflow, labor, regulation or hardware complexity. Yugabyte note · Fireworks note · Multiverse note
• Mechanism and reasoning — Enablers can scale across customers with lower domain burden but risk commoditization/bundling. Full-stack operators can capture outcome value and proprietary operations but take distribution, regulation, services, working-capital or capex risk. [researcher inference · strong]
• Historical and vehicle context — Software-era Lightspeed cases favored reusable enterprise layers; the 91 cards also include education, lending, robotics, health, defense and energy operators. Check size, reserve and specialist governance must fit the chosen layer. Ledger
• When to use — Category formation where both infrastructure and vertical operators are plausible. Do not force the taxonomy when the company is a consumer network or a science asset without an operating workflow.
• How to apply — Map who owns customer outcome, distribution, data, regulatory liability and capex; compare gross margin/sales cycle/integration; model incumbent bundling versus vertical expansion; choose vehicle/reserves; kill a hybrid that inherits both risk sets without control.
• Examples —
  • Yugabyte — researcher-applied enabler illustration: a distributed-database layer seeks reuse across customers; incumbent clouds and operational complexity are counterweights. Yugabyte note
  • Glean — researcher-applied enabler illustration: the knowledge/control layer seeks workflow breadth, while incumbent bundling and permission depth remain key tests. Glean note
  • Fireworks — investor-stated current enabler thesis; outcome immature: performance, margin and model-provider dependence are unresolved. Fireworks note
  • Blend — researcher-applied full-stack illustration: a lending operator takes more workflow and regulatory execution than a generic software layer. Blend note
  • Multiverse — researcher-applied full-stack illustration: apprenticeship/outcome delivery adds labor, employer and education execution. Multiverse note
  • Dexterity — researcher-applied full-stack illustration: robotics absorbs hardware/integration and physical-world reliability risk. Dexterity note
  • Hybrid — hypothetical boundary: a vendor selling generic infrastructure while promising vertical outcomes can inherit bundling plus services/capex risk; no matched Mhatre portfolio cohort proves the winning boundary. Fireworks/Blend premise controls [researcher inference · moderate]
• Evidence — The rule is triangulated from independent company categories and direct/interested works: Yugabyte, Blend filing context, Multiverse and Dexterity. These show distinct operating models, not Mhatre's explicit general doctrine.
• Limits, failure modes, and exception ledger — Enablers can be bundled; full-stack companies can have superior data/distribution; category boundaries move; capex/regulation can create defenses; website labels omit economics. Log control point, payer, liability, gross margin, capex, regulation, incumbent response and vehicle fit.
• Evolution and contradictions — Enterprise infrastructure dominates the earlier teaching record while later cards widen to vertical labor, finance, robotics, health and frontier science. The model is therefore a provisional allocator, not a quotation or observed IC rule. [researcher inference · moderate]
• Modern VC translation (as of 2026-08-23) — Fireworks is the grounded current enabling example and Temporal a workflow-control layer; current vertical AI operators remain too immature for a matched outcome claim. Test whether the startup owns durable workflow value or merely resells model capability, and size capital to the chosen layer. Fireworks note · Temporal note [researcher inference · moderate]

Cross-model decision worksheet

1. Freeze the date, adviser, vehicle, source owner and current “gravity.”
2. Write the technical/local change, buyer constraint and exact wedge; choose enabler versus full-stack economics.
3. Separate relationship evidence from the outside-view market/product/price case.
4. Define zero-to-one, expansion, trust/reliability and independent-readiness gates.
5. Map founder/team capability, firm-function owners, permissions/context and company feedback loop.
6. Model security, ownership, dilution, reserves and downside by vehicle; do not use company valuation as return.
7. Prewrite the observation that changes conviction, capital, governance or stop decision.
8. Re-underwrite on evidence or regime change and preserve dissent, exceptions and unresolved attribution.

Exception ledger and anti-hagiography controls

• A selected winner essay is rationale evidence after success, not an original memo.
• A firm card is association evidence, not personal source, lead, board, ownership or return.
• A founder account is a retrospective/relational witness unless independently timed.
• A filing is strong for a dated security/board snapshot, not rationale, cash realization or sole decision rights.
• BYJU'S is an adviser/team adverse association control; it does not prove Mhatre causation or misconduct.
• Recent AI/deep-tech cases ground current translations but cannot validate outcomes.
• Stanford's two science errors require domain correction and keep technical fluency separate from factual authority. Talk control

Excluded slogans and candidate theses

“Technology is a language,” “moon/rocket,” “crazy goals,” “controlled recklessness,” “every company is a tech company,” “interesting times,” “student-athlete” and “Move 37” are memorable vocabulary, not standalone decision models; their useful mechanisms are absorbed above. Jobs-not-walls and health convergence are policy/sector positions, not sufficiently repeated operating rules.

RaaS, employer-paid/outcome pricing, sovereign AI, multiple-model winners and training/inference amortization remain candidate theses. Each lacks three grounded Mhatre applications with independent outcome/economic evidence; promote one only after a dated thesis, three decision records, countercase and falsifier are found. [researcher inference · insufficient public record]

Suggested study path

Start with Models 1–3 for architecture, wedge and capital staging; add Models 4, 8 and 7 for people/access/team attribution; use Models 5–6 for updating; apply Models 10–12 for trust, context and geography; finish with Models 9, 13 and 14 for readiness, security economics and value-layer choice. Then complete all seven decide-before-reveal labs.

Evidence gaps and candidate models

The excluded slogans and candidate theses above remain part of this inventory: RaaS, employer-paid/outcome pricing, sovereign AI, multiple-model winners and training/inference amortization lack a dated rule plus three grounded decisions, an independent outcome/control and a falsifier. The decisive retrieval targets are original dated memos; complete opportunity/pass/write-off cohorts; customer/reference and dissent logs; partner-function and board/recruiting records; each follow-on's milestone memo; private security/term/reserve histories; incident/retention/unit-economics data; and audited vehicle cash flows. Those gaps cap the combined edge at moderate even when individual direct rules are strongly evidenced. [researcher inference · strong]

What the evidence supports

The record supports a narrower and more useful thesis than “Ravi Mhatre picked many winners.” Across the Stanford process talk, especially 07:00–15:05 and 26:28–31:40, the MuleSoft founder account, the AppDynamics S-1, “Principal Stockholders” and director biography, and the Rubrik 2026 proxy, Mhatre repeatedly appears where a technical inflection can be translated into a narrow enterprise wedge, an adaptable founder can recruit around a long ambition, and a firm platform can support later stages. [researcher inference · moderate] The component rules are directly voiced or behaviorally documented; their combined effectiveness is only moderate because original memos, rejected opportunities, milestone gates, complete follow-on records and vehicle cash flows are absent.

The evidence is strongest for method and role boundaries, not personal performance. Seven ex-ante labs preserve what was known and unknown before outcome disclosure; six are major enterprise cases and one is the BYJU'S failure/adverse control. The current official profile exposes 91 company cards, but a card proves association only. It does not prove sourcing, lead, board work, vehicle, ownership or economics. The portfolio denominator therefore reports 91/91 card-level coverage, 7/91 deep reconstruction and a lifetime opportunity/pass/write-off coverage fraction that is not computable. [documented behavior · strong]

Three recurring conclusions survive adversarial reopening:

• Technical architecture matters only when it resolves a named buyer or operating constraint. The contemporaneous Zscaler 2012 financing release, Materialize thesis and Yugabyte thesis support the rule; broad crypto and recent frontier-AI forecasts remain weaker tests. [investor-stated · strong]
• Platform ambition is earned through a hard wedge, founder adaptation and staged proof. MuleSoft, AppDynamics, Rubrik and Glean support the pattern, but their public accounts are success-selected and often retrospective. [researcher inference · moderate]
• Team and vehicle attribution are part of the investment result, not footnotes. The Lightspeed legal page separates LSVP, Lightspeed India Partners LLP and Faction; issuer filings separate numbered funds, SPVs, shared general-partner direction and personal/family lines. [documented behavior · strong]

The biography also survives only in source-bounded form. Stanford's alumni directory supports the 1989 AB and 1993 MBA years; the 2020 Stanford transcript, pp.3–4 supports the degree types and family-profession memory, not those years. The 2000 WPG/Lightspeed report names six founding general partners and institutional continuity; Mhatre's later four-classmate recollection in TiEcon 2017, 02:29–04:08 and IIT Bay Area 2017, 18:14–20:33 describes a narrower core group. Both remain visible rather than being smoothed into one founding story. [researcher inference · strong]

What could falsify the thesis

The favorable thesis is four separable hypotheses, not one halo effect:

The strongest skeptical explanation is access plus survivorship: an established multi-stage platform obtains repeat-founder introductions, can reserve into winners, and publishes narratives after success. That can produce the observed record without a unique picking edge. The ThoughtSpot chronology conflict, the lack of a Zscaler lead/board record, the pre-IPO AppDynamics sale, firm-wide crypto forecasting, and BYJU'S association/adverse boundary prevent a clean hero narrative. [researcher inference · strong] Discriminating evidence would be dated memos, the full opportunity/pass cohort, milestone-specific partner actions, and audited vehicle cash flows excluding the top outcomes.

BYJU'S is a genuine failure/governance exercise but not a clean personal falsifier. Its official card shows Mhatre association across distinct LSVP/LSIP events; the issuer investor page separately identifies Dev Khare/LSIP. The IBBI process and Supreme Court judgment establish procedural outcome, not Mhatre lead, board role, stake, loss, causation or misconduct. [contemporaneous record · strong]

Attribution and denominator audit

The mechanical reconciliation is exact at the frozen 2026-08-23 boundary:

The 16 rows that entered REVIEW unchecked were not mass-waived. Each was re-opened or its named search route re-run, then classified:

High-risk claim reopening trail

Transfer limits for modern VCs

Fourteen modern-transfer tests preserve the mechanism, changed assumptions, signals, likely misuse and current example for each model. Architecture-to-wedge, staged proof, recoverability, permissioned context and outside-view controls transfer best. The Temporal durability thesis, Anthropic safety thesis, Fireworks inference thesis and Ricursive loop thesis are grounded current examples, but they are interested, recent and outcome-immature. [researcher inference · moderate]

The least portable elements are outcome-derived confidence, unrestricted reserve capacity, repeat-founder access and assumptions formed in enterprise software's long 2007–2021 expansion. A modern analyst should require dated adoption evidence, security-specific ownership, explicit follow-on gates and vehicle fit; current AI valuations, adviser RAUM and company importance are not returns. Vehicle economics [researcher inference · strong]

The models are decision prompts, not causal laws. Model 6's “gravity” re-underwriting is investor-stated as a regime concept, while applying it to AppDynamics or Rubrik is researcher inference. Models 11–14—enterprise context, geography/entity fit, company-versus-security separation and enabling-layer-versus-full-stack choice—have useful cases but thinner proof of being actual Mhatre decision rules. Evidence gaps and candidate models [researcher inference · moderate]

Unresolved evidence

No public record supplies the complete opportunity/pass/write-off universe, original IC memos, partner votes, board work logs, recruiting attribution, milestone gates, follow-on reserve schedule, security-by-security cost/dilution/distribution history, fund cash flows, carry allocation, LP communications or audited Mhatre-specific returns. The package therefore cannot compute selection accuracy, loss rate, ownership-adjusted contribution, realized return, personal alpha or performance excluding dominant outliers. Source gaps [researcher inference · insufficient public record]

The one spoken-access gap is Bloomberg's 2016 Nutanix video: official metadata resolves, but no transcript, caption track or legitimately retrievable audio was found. Talk note Medium's author index is intermittently accessible, so 40 writings are a bounded selected corpus, not an exhaustive lifetime catalog. SearchWorks and the Computer History Museum returned no dedicated Mhatre archive/oral history. These are precise exhausted-route boundaries, not permission to reconstruct missing speech or documents. [researcher inference · insufficient public record]

The highest-value future evidence is specific: dated seed/Series A memos for MuleSoft, AppDynamics, Nutanix and Rubrik; a complete opportunity/pass cohort; board/recruiting/follow-on logs tied to milestones; fund-by-fund cash flows and distributions; and an official transcript for the 2016 Bloomberg appearance. Any of those could change the edge grade, allocation of team credit or economic ranking. Until then, the central edge stays [researcher inference · moderate], not strong.

Reproducible ordinary-claim audit trail

This sample was fixed before checking results: two ordinary claims from each of the seven pre-REVIEW roots, a literal distribution of 2/2/2/2/2/2/2. It is separate from the high-risk census above.

Twenty-point acceptance evidence

Direct writings and firm-authored record

• ★★ | If Every Company Is a Tech Company, What Do Tech Companies Do? | https://medium.com/lightspeed-venture-partners/if-every-company-is-a-tech-company-what-do-tech-companies-do-32c759292b01 | 2016-10-25 | direct company-building thesis; reliability A, full | early organizational/technology doctrine and contemporary context
• ★★ | What Do You Call a Tech Unicorn That Goes Public? | https://medium.com/lightspeed-venture-partners/what-do-you-call-a-tech-unicorn-that-goes-public-1cbe02987932 | 2016-09-19 | direct enterprise-IPO essay; reliability A, full | early public-readiness doctrine and market-cycle boundary
• ★★ | IPOs Head to the Enterprise Cloud | https://medium.com/lightspeed-venture-partners/ipos-head-to-the-enterprise-cloud-49afc846f43f | 2016-12-20 | direct enterprise-market essay; reliability A, full | cloud cohort and exit context; forecast/outcome separation required
• ★★★ | Purpose Built: Nutanix | https://medium.com/lightspeed-venture-partners/purpose-built-if-you-dont-engineer-your-rocket-ship-to-go-to-the-moon-it-never-will-f1d25e7f298 | 2016-09-30 | direct essay; reliability A, full | public-company readiness, global-from-day-one scaling and mixed-experience teams
• ★★ | Building Jobs, Not Walls | https://medium.com/lightspeed-venture-partners/building-jobs-not-walls-in-silicon-valley-fdcf7fc15d2d | 2016-11-24 | direct essay, originally TechCrunch; reliability A, full | family history, immigration and greenfield job-creation thesis
• ★★★ | Building a Multi-Billion Company the AppDynamics Way | https://medium.com/lightspeed-venture-partners/building-a-multi-billion-company-the-appdynamics-way-f97dca401509 | 2017-01-25 | direct retrospective; reliability A, full | extreme-purpose founder thesis and Lightspeed/team attribution before Cisco sale
• ★★★ | Doubling Down — Our Investment in Rubrik | https://medium.com/lightspeed-venture-partners/doubling-down-5c9e7e8460e0 | 2017-05-09 | direct deal interview; reliability A, full | incubation, follow-on conviction and reported $140m firm commitment; not personal economics
• ★★ | Predicting Enterprise Trends in 2018 | https://medium.com/lightspeed-venture-partners/predicting-enterprise-trends-in-2018-d201fa74513a | 2018-01-16 | direct outlook; reliability A, full | hybrid cloud, security and physical-world technology forecast; permits forecast audit
• ★★ | India's Mobile Internet Revolution and ShareChat's $100M Series C | https://medium.com/lightspeed-venture-partners/indias-mobile-internet-revolution-and-sharechat-s-100m-series-c-1e598d5e540f | 2018-09-27 | Mhatre/Dev Khare coauthored investment essay; reliability A, full | India round and market thesis; LSIP/entity and partner credit must be preserved
• ★★ | Zscaler: Innovation and Audacity Win the Race | https://medium.com/lightspeed-venture-partners/zscaler-innovation-and-audacity-win-the-race-4ff6e7020f95 | 2018-03-16 | direct retrospective; reliability A, full | founder audacity and security-category evidence; cross-check team credit
• ★★ | The Future of Business Intelligence | https://medium.com/lightspeed-venture-partners/the-future-of-business-intelligence-55c69bd5a2da | 2019-08-28 | Mhatre/Amy Wu essay; reliability A, full | repeat-founder network and ThoughtSpot follow-on; coauthor attribution required
• ★★ | Enterprise CEO Advisory Council | https://medium.com/lightspeed-venture-partners/announcing-the-lightspeed-enterprise-ceo-advisory-council-f8225447e1ad | 2019-10-02 | direct institutional essay; reliability A, full | operating-adviser platform and named recruiting/mentoring roles
• ★★ | Next-Level Defense vs Next-Gen Security Threats / Confluera | https://medium.com/lightspeed-venture-partners/next-level-defense-vs-next-gen-security-threats-6c9044208066 | 2019-07-30 | direct investment essay; reliability A, full | security-market thesis and explicit Bipul Sinha/Abhijit Ghosh/Niloy Mukherjee roles
• ★★ | Aurora: Powering the Next Revolution in Transportation | https://medium.com/lightspeed-venture-partners/aurora-powering-the-next-revolution-in-transportation-10e8a3c2c749 | 2019-02-08 | direct/coauthored investment essay; reliability A, full | ex-ante autonomous-transport thesis and partner attribution
• ★★ | Growing Our Team at Lightspeed | https://medium.com/lightspeed-venture-partners/growing-our-team-at-lightspeed-c531256f2136 | 2020-07-21 | direct institutional essay; reliability A, full | diversity, specialization and executive-talent platform design
• ★★ | Enabling a Blameless Culture | https://lsvp.com/stories/enabling-a-blameless-culture/ | 2019-03-20 | official body signed Ravi Mhatre/Jay Madheswaran; Medium metadata is incomplete; reliability A, full | learning culture, follow-on and organizational mechanism; body co-signature governs attribution
• ★★ | Materialize: Streaming Without Compromises | https://medium.com/lightspeed-venture-partners/materialize-streaming-without-compromises-b85d6d520532 | 2020-11-30 | direct investment essay; reliability A, full | database architecture and technical-market thesis
• ★★ | YugabyteDB: Distributed SQL Has Arrived | https://medium.com/lightspeed-venture-partners/yugabytedb-distributed-sql-has-arrived-204796e55ca1 | 2021-03-03 | direct investment essay; reliability A, full | cloud-native distributed SQL thesis and technical-founder evidence
• ★★ | Dexterity: Intelligent Robots | https://medium.com/lightspeed-venture-partners/dexterity-ushering-a-new-era-of-intelligent-robots-3b80cbf57167 | 2020-07-21 | direct investment essay; reliability A, full | computing-to-physical-world thesis and founder/technology evidence
• ★★ | Health Convergence | https://lsvp.com/stories/the-convergence-of-technology-and-health-making-lives-better-for-patients/ | 2021-06-29 | Mhatre/Kohler/MacQuitty essay; reliability A, full | team-led sector build, vehicle flexibility and explicit Ling Wong credit
• ★★ | Better-Lending Platform Blend Goes Public | https://medium.com/lightspeed-venture-partners/better-lending-platform-blend-goes-public-65712ec89612 | 2021-07-16 | direct IPO essay; reliability A, full | founder-first claim and firm fintech denominator; person-level attribution not established
• ★★★ | Offchain Labs: Arbitrum | https://medium.com/lightspeed-venture-partners/offchain-labs-how-arbitrum-is-transforming-ethereum-scalability-510685b0a159 | 2021-08-31 | Amy Wu/Mhatre investment essay; reliability A, full | technical scalability thesis with coauthor/team control
• ★★★ | Glean: Bringing Search to the Workplace | https://medium.com/lightspeed-venture-partners/glean-bringing-search-to-the-workplace-f8fcfef5175c | 2021-09-15 | Mhatre/Ansaf Kareem essay; reliability A, full | repeat-founder sourcing, initial co-lead and quantified customer claims
• ★★ | Driving Innovation: Aurora Goes Public | https://medium.com/lightspeed-venture-partners/driving-innovation-aurora-goes-public-5605378ea4db | 2021-11-04 | Raviraj Jain/Mhatre essay; reliability A, full | team/partnership/technology underwriting; market-size figure requires independent check
• ★★★ | Investing in Alchemy | https://medium.com/lightspeed-venture-partners/investing-in-alchemy-ad8a127314c3 | 2022-02-08 | direct investment essay; reliability A, full | developer-platform pattern and blockchain analogy; live-mark caveat
• ★★ | The Future of Enterprise Search: Glean | https://medium.com/lightspeed-venture-partners/the-future-of-enterprise-search-glean-3eec4fef8956 | 2022-05-18 | Mhatre/Ephrati/Kareem essay; reliability A, full | initial, Series B and Series C chronology plus lifecycle investing
• ★★ | Building Blockchain Infrastructure / Anand Iyer | https://medium.com/lightspeed-venture-partners/building-blockchain-infrastructure-one-startup-at-a-time-meet-venture-partner-anand-iyer-a2bdafb25d56 | 2022-07-20 | direct institutional essay; reliability A, full | reported firm-wide crypto capital and personnel, not personal portfolio evidence
• ★★ | Lightspeed Adds Experienced Growth Investor | https://medium.com/lightspeed-venture-partners/https-medium-com-ravi-lsvp-lightspeedbradleytwohig-644ddcbc3d56 | 2018-07-10 | multi-author institutional essay; reliability A, full | early/growth lifecycle vehicle design and reported $1.8b fund family; not Mhatre-only doctrine
• ★★ | The Future of Higher Education: Multiverse | https://lsvp.com/stories/the-future-of-higher-education-multiverse/ | 2022-06-08 | official body authors Mandan/Ephrati/Quinn/Mhatre; reliability A, full | labor/education application thesis; Mandan lead/structured author
• ★★★ | Typeface: Personalized Generative AI | https://medium.com/lightspeed-venture-partners/typeface-transforming-enterprise-content-creation-with-radically-personalized-generative-ai-ef6e857e49aa | 2023-02-27 | Mhatre/Raviraj Jain essay, co-created with company and AI-assisted | early generative-AI thesis with explicit marketing/authorship boundary
• ★★ | Glean's Enterprise-Ready Generative AI | https://lsvp.com/stories/gleans-enterprise-ready-generative-ai-is-transforming-workplace-search-and-discovery/ | 2023-05-01 Medium / 2023-05-02 migration | direct essay; reliability A, full | trusted enterprise knowledge as AI moat; date/canonical migration preserved
• ★★ | Deepening Our Partnership with Anthropic | https://lsvp.com/stories/deepening-our-partnership-with-anthropic/ | 2025-09-02 | Mhatre/Chahal/Duesterhoeft firm investment essay; reliability A, full | safe frontier-model thesis and Series F; live-mark/economics boundary
• ★★ | Doubling Down on Descope | https://lsvp.com/stories/doubling-down-on-lightspeeds-investment-in-descope-the-next-gen-iam-platform-for-customers-partners-and-ai-agents/ | 2025-09-30 | Chahal/Mhatre investment essay; reliability A, full | identity infrastructure across human and agent users; team and follow-on attribution required
• ★★ | Unconventional AI: Biology-Scale Efficiency | https://lsvp.com/stories/investing-in-unconventional-ai-biology-scale-efficiency-for-the-ai-era/ | 2025-12-08 | Chahal/Mhatre/Moore investment essay; reliability A, full | efficiency/architecture thesis and research-team diligence; live-mark caveat
• ★★ | AI Safety vs AI Scalability Is a False Dichotomy | https://lsvp.com/stories/ai-safety-vs-ai-scalability-is-a-false-dichotomy/ | 2025-10-27 | Mhatre sole-author thesis; reliability A, full | safety/scaling complementarity and enterprise-trust thesis
• ★★ | Backing America's AI Leadership | https://lsvp.com/stories/backing-americas-ai-leadership-applauding-the-white-house-ai-action-plan/ | 2025-07-23 | firm policy essay with direct Mhatre quotation and no Authors block; reliability A/C, full | national-policy/energy thesis; institutional authorship boundary explicit
• ★★ | Fireworks AI: Inference Platform | https://lsvp.com/stories/our-investment-in-fireworks-ai-the-inference-platform-aiming-to-power-every-genai-application/ | 2025-10-28 | Vaswani/Mhatre/Duesterhoeft/Chahal/Shen essay; reliability A, full | infrastructure-layer pattern; five-author team credit required
• ★★ | Ricursive Intelligence | https://lsvp.com/stories/investing-in-ricursive-intelligence-ai-for-chip-design-and-chip-design-for-ai/ | 2026-01-26 | Chahal/Mhatre/Cader essay; reliability A, full | prior-work diligence, technical depth and chip-design loop
• ★★ | Temporal: Durable Agents | https://lsvp.com/stories/announcing-our-investment-in-temporal-agents-make-mistakes-your-workflows-cant/ | 2026-02-17 | Vaswani/Mhatre/Moore/Shen essay; reliability A, full | observed customer-pattern sourcing and reliability infrastructure
• ★★★ | Move 37, Ten Years On: Ineffable Intelligence | https://lsvp.com/stories/move-37-ten-years-on-partnering-with-ineffable-intelligence/ | 2026-04-27 | Mhatre/Jain/Moyroud essay; reliability A, full | experiential-learning thesis, research pedigree and safety disposition

Talks, interviews, podcasts and video

• ★★★ | TiEcon interview | https://www.youtube.com/watch?v=S79P83FMZ0U | 2012-10-03 | 14:32 provider-captioned interview; reliability A/B, full | family/SGI/Netscape/1995 venture-entry chronology; BPcxfbezsvA excerpt deduped
• ★★★ | Nutanix at .NEXT / theCUBE | https://www.youtube.com/watch?v=uehI2YhYc0g | 2015-06-09 | 23:34 provider-captioned interview; reliability A/B, full | architecture, simplicity and hybrid-cloud transition; company-event selection bias
• ★★★ | Identifying a Billion-Dollar Company at Idea Stage | https://www.geektime.co.il/an-interview-with-ravi-mhatre/ | 2015-11-30 | Hebrew edited interview; reliability A/B, translated full text | lean 0→1/1→∞, ThoughtSpot and claimed one-in-100 funnel; translation/denominator caveat
• ★★ | CNBC-TV18: India investment | https://www.youtube.com/watch?v=sBmP-MjSS2s | 2015-09-26 | 12:44 provider-captioned interview; reliability A/B, full | 2005 India-office memory, smartphone/data thesis and period valuation claim
• ★★ | Bloomberg: Nutanix IPO reception | https://www.bloomberg.com/news/videos/2016-10-07/what-the-positive-reception-of-nutanix-means-for-tech-ipos | 2016-10-07 | official embedded video page; reliability A/B, partial metadata only | no transcript/captions or legitimate audio route recovered; no speech inferred
• ★★★ | Lightspeed Cofounders' 17-Year Overnight Success | https://www.youtube.com/watch?v=3V_y7Y7xx9U + https://techcrunch.com/2017/05/10/lightspeeds-cofounders-on-their-17-year-long-overnight-success-story/ | 2017-05-10 | 15:02 captions plus edited transcript; reliability A/B, full/edited | enterprise-first platform, early AppDynamics and capital-compression thesis
• ★★★ | TiEcon keynote interview | https://www.youtube.com/watch?v=psBR-vy8Vzk | 2017-05-16 | 22:48 provider-captioned interview; reliability A/B, full | formation memory, crash adversity and founder ambition
• ★★ | IIT Bay Area 2017 appearance family | https://www.youtube.com/watch?v=BF9UGJHjP7k + https://www.youtube.com/watch?v=F7q5U4gQBV8 | 2017-06-22 / 2017-07-29 | complementary captioned event assets; reliability A/B, full | family/company chronology and unique four-founder/technical-founder window; one family
• ★★★ | AppDynamics founder journey | https://www.youtube.com/watch?v=N2mUVXkw41w | 2018-06-04 | 19:45 no-caption video; OpenAI whisper-1 segment transcription; reliability B, full audio | phase-specific company building and founder adaptation; no transcript committed
• ★ | On Setting Crazy Goals and Running Toward Them | https://lsvp.com/event/on-setting-crazy-goals-and-running-toward-them/ | 2018-07-11 | event context/images; reliability A/C, partial | convening behavior only; no speech or individual contribution inferred
• ★★★ | MuleSoft founder–investor fireside | https://www.youtube.com/watch?v=_AO2_Uzpdz4 | 2018-09-07 upload | 34:40 captioned excerpt; reliability A/B, full excerpt | Greg Schott/Mhatre speaker map; LM7kZsSRNBM longer asset deduped
• ★★★ | Founder to CEO | https://inc42.com/resources/the-journey-from-a-founder-to-a-ceo-with-ravi-mhatre/ | 2018-10-15 | moderator notes/edited excerpts; reliability B, full text | founder reinvention, learning and superpower diagnostics; not verbatim
• ★★★ | The Language of Technology | https://www.youtube.com/watch?v=1uwzycO_E8c + https://ecorner.stanford.edu/wp-content/uploads/sites/2/2020/11/the-language-of-technology-entire-talk-transcript.pdf | 2020-11-05 / 2020-11-11 | 45:10 video and official timestamped transcript; reliability A, full | richest process source; scientific asides externally corrected
• ★★ | Bloomberg: future of crypto | https://www.youtube.com/watch?v=TXeGOCISrVg | 2021-12-10 | 4:55 no-caption video; OpenAI whisper-1 segment transcription; reliability B, full audio | regulation balance, firm crypto history and forecast countercontrol
• ★★ | Snorkel AI fireside/product session | https://www.youtube.com/watch?v=SsGbBTBff9A | 2022-04-27 | 1:01:46 provider-captioned multi-speaker session; reliability A/C, full | Ratner product claims separated from Mhatre participation
• ★★★ | Venture Capital Gravitational Physics | https://www.youtube.com/watch?v=U8VLDyrj6Mk + https://www.newcomer.co/p/venture-capital-gravitational-physics | 2023-03-29 | 43:15 captioned interview/podcast; reliability A/B, full | post-SVB gravity/regime model; publisher has no transcript
• ★★ | Worth/Milken: AI market impacts | https://www.youtube.com/watch?v=oKZrrlK4oew | 2024-05-20 | 4:02 provider-captioned interview; reliability A/B, full | public AI inflection, stack value capture and peak-capital forecast
• ★★ | BCG: adapting to generative AI | https://www.youtube.com/watch?v=t-h4NiKH7PM | 2024-05-10 | 2:48 manual/automatic-captioned interview; reliability A/B, full | people, markets and informed early-lead criteria
• ★★ | Stanford ZKP/blockchain panel | https://www.youtube.com/watch?v=FAVz5IjyWks | 2024-07-12 | 51:10 manual/automatic-captioned panel; reliability A/B, full | 2013 firm history and architectural-platform thesis; panelists separated
• ★★ | NYSE Interesting Times, Episode 4 | https://www.youtube.com/watch?v=X90SGz1BivU | 2024-09-10 | 7:30 no-caption video; OpenAI whisper-1 segment transcription; reliability B, full audio | AI realization, misinformation/security and investor responsibility; i85E4mJuWYE deduped
• ★★★ | Cheddar: Rubrik IPO | https://www.youtube.com/watch?v=fCmLZQ7FJ-A | 2024-04-25 | 4:51 provider-captioned interview; reliability A/B, full | platform broadening, recoverability and public-company stewardship
• ★★ | Axios View From the Top: enterprise AI ROI | https://www.axios.com/2024/12/19/axios-event-ai-investment-glean-ceo | 2024-12-16 event / 2024-12-19 report | one-sentence sponsored-event readout; reliability B, partial | experiment-to-ROI milestone; separate from Arvind Jain interview
• ★★★ | CNBC: AI valuations and Anthropic | https://www.youtube.com/watch?v=QocqJDTjGvA | 2025-03-14 | 6:35 manual/automatic-captioned interview; reliability A/B, full | training/inference economics and multi-winner forecast
• ★★ | Hill & Valley Forum: AI and the American Fabric | https://www.youtube.com/watch?v=SsEoFDc1DIg | 2025-05-03 | 26:55 provider-captioned panel; reliability A/B, full | Mhatre moderator; Rounds/Clark policy claims separated; Apple mirror deduped
• ★★★ | Jefferies: VC's Role in the Age of AI | https://www.jefferies.com/insights/markets/ravi-mhatre-on-venture-capitals-role-in-the-age-of-ai/ | 2025-06-02 | edited direct interview; reliability A/B, full | expertise amplification, institutional venture and buy-forward doctrine
• ★★★ | ETtech: AI Bubble and India | https://economictimes.indiatimes.com/tech/technology/ettech-interview-lightspeeds-ravi-mahtre-bejul-somaia-on-the-ai-bubble-india-opportunity-more/articleshow/125419650.cms | 2025-11-19 | edited Mhatre/Somaia interview; reliability A/B, full | bubble/transformation, usage/talent and sovereign AI; speaker/entity boundary
• ★★★ | Bloomberg: AI moves from demos to deployment | https://www.youtube.com/watch?v=bDsVAE9cwmk | 2026-02-25 | 10:21 manual/automatic-captioned interview; reliability A/B, full | trust/reliability and firm-wide 165-company host premise; DailyMotion mirror deduped
• ★★★ | CNBC International: software reborn | https://www.youtube.com/watch?v=i3pdnKIoUC0 | 2026-02-18 | 6:54 provider-captioned interview; reliability A/B, full | agents, context/data moat and Anthropic thesis
• ★★ | DualEntry: future of ERP | https://www.youtube.com/watch?v=zcgMBdQMC-8 | 2026-02-05 | 1:44 manual/automatic-captioned montage; reliability A/C, full | ERP-specific AI depth; unlabeled-speaker limits
• ★★★ | Lightwork: thinking in future tense | https://www.youtube.com/watch?v=M2NHdPYA8z0 | 2026-06-17 | 16:09 provider-captioned firm interview; reliability A/C, full | moving-frontier founder test and institutional marketing boundary
• ★★ | Film Session: student-athlete lessons for venture | https://www.youtube.com/watch?v=cjeFOc5MolI | 2026-08-13 | 4:00 provider-captioned excerpt; reliability A/B, full | adversity, differentiated team roles and analogy limit

Firm, filings, company and deal evidence

• ★★★ | Official Ravi Mhatre profile | https://lsvp.com/team-member/ravi-mhatre/ | current; accessed 2026-08-23 | official biography/portfolio route; reliability A/C, full | 10 named-company portfolio anchor; image-linked investments require recovery; direct maxims, current title and entity disclaimers
• ★★★ | Stanford alumni directory | https://alumni.stanford.edu/stanford-associates/directory/ | current; accessed 2026-08-23 | university record; reliability A, full | class of 1989 and MBA 1993 identity control
• ★★★ | Weiss Peck & Greer Becomes Lightspeed | https://www.lightreading.com/cable-technology/weiss-peck-greer-becomes-lightspeed | 2000-10-11 | contemporaneous trade press; reliability B, full | WPG continuity, six founding GPs and reported $1b vehicle
• ★★ | India Office Announcement | https://economictimes.indiatimes.com/news/economy/finance/lightspeed-venture-to-set-shop-in-india/articleshow/2047642.cms | 2006 | contemporaneous press/direct quote; reliability A/B, full | Mhatre's period globalization rationale; later LSIP entity controls apply
• ★★★ | AppDynamics S-1 | https://www.sec.gov/Archives/edgar/data/1435043/000119312516805559/d209425ds1.htm | 2016-12-28 | issuer filing; reliability A, full | Mhatre director since April 2008; Lightspeed VII 22,677,444 shares/20.8% in proposed-IPO/pre-offering snapshot under shared power; later-round slices are not total basis
• ★★★ | Cisco completes AppDynamics acquisition | https://newsroom.cisco.com/c/r/newsroom/en/us/a/y2017/m03/cisco-completes-acquisition-of-appdynamics.html | 2017-03-22 | acquirer announcement; reliability A/C, full | approximately $3.7b cash plus assumed equity awards; issuer transaction value, not Lightspeed proceeds or return
• ★★★ | Nutanix final prospectus | https://www.sec.gov/Archives/edgar/data/1618732/000119312516727084/d937439d424b4.htm | 2016-09-30 | issuer filing; reliability A, full | Mhatre director July 2010 and lead independent August 2015; Lightspeed affiliate 22.7% pre/20.3% post under shared power
• ★★ | Nutanix Form 4 | https://ir.nutanix.com/static-files/f9e38967-48c7-4ad8-971d-e81f68965934 | 2017-06-09 | insider filing; reliability A, full | in-kind distribution of 117,188 shares into Mhatre Investments LP Fund 1, leaving 206,350 shares after; four UGP directors share direction and disclaim except pecuniary interest, so this is not a personal-return record
• ★★★ | MuleSoft final prospectus | https://www.sec.gov/Archives/edgar/data/1374684/000119312517086142/d287291d424b4.htm | 2017-03-17 | issuer filing; reliability A, full | Mhatre director May 2007; Lightspeed VII/Select 17.0% pre/15.3% post under shared power
• ★★ | Ross Mason official profile | https://lsvp.com/team-member/ross-mason/ | current; accessed 2026-08-23 | official founder account; reliability A/B, full | Mhatre-led initial/subsequent MuleSoft investment claim and founder perspective
• ★★★ | Salesforce–MuleSoft definitive agreement | https://www.salesforce.com/news/press-releases/2018/03/20/salesforce-signs-definitive-agreement-to-acquire-mulesoft/ | 2018-03-20 | acquirer announcement; reliability A/C, full | approximately $6.5b enterprise value; not equity proceeds, vehicle proceeds or return
• ★★★ | MuleSoft acquisition close 8-K | https://www.sec.gov/Archives/edgar/data/1374684/000119312518147738/d571994d8k.htm | 2018-05-02 | issuer filing; reliability A, full | transaction close control; no allocation of proceeds to Lightspeed or Mhatre
• ★★ | Zscaler $38m Investment | https://www.zscaler.com/press/zscaler-secures-38-million-investment-round | 2012 | company release/direct Mhatre quote; reliability A/C, full | contemporaneous security thesis and Lightspeed participation
• ★★★ | Zscaler S-1/A | https://www.sec.gov/Archives/edgar/data/1713683/000119312518080170/d400527ds1a.htm | 2018-03-12 | issuer filing; reliability A, full | founder trust and TPG are disclosed 5%+ holders; Lightspeed/Mhatre absence above threshold is not proof of zero
• ★★★ | Rubrik 2024 S-1/A | https://www.sec.gov/Archives/edgar/data/1943896/000119312524097005/d359771ds1a.htm | 2024-04-16 | issuer filing, “Principal Stockholders,” p.169; reliability A, full | Lightspeed 38,984,426 Class B shares/23.9% before offering and 38,984,426/25.6% of Class B/25.4% voting power after; dated snapshot kept separate from the 2026 proxy
• ★★★ | Rubrik 2026 Proxy | https://www.sec.gov/Archives/edgar/data/1943896/000194389626000034/rbrk-20260415.htm | 2026-04-15 | issuer filing; reliability A, full | current board date, Lightspeed holdings, shared voting/dispositive-power boundary
• ★★ | ThoughtSpot current team/board record | https://www.thoughtspot.com/team | current; accessed 2026-08-23 | company leadership route; reliability A/C, full | current board snapshot; distinct from historical 2014 board evidence
• ★★★ | ThoughtSpot $10.7m Series A announcement | https://www.thoughtspot.com/press-releases/don%E2%80%99t-bi-just-search-thoughtspot-brings-google-user-experience-business-data | 2014-02-05 | issuer announcement; reliability A/C, full | firm-led Series A and 2014 Mhatre board; distinct from current board and card's 2012 Seed metadata
• ★★ | ThoughtSpot Series F announcement | https://www.thoughtspot.com/press-releases/thoughtspot-raises-new-funding-at-usd4-2b-valuation | 2021-11-15 | issuer announcement; reliability A/C, full | March Capital-led financing and reported $4.2b valuation; not proceeds or realized return
• ★★ | BYJU'S official investor page | https://byjus.com/our-investors/ | current; accessed 2026-08-23 | issuer relationship page; reliability A/C, full | separately identifies Dev Khare/LSIP relationship; does not allocate Mhatre lead, board, stake or loss
• ★★ | Natera S-1 | https://www.sec.gov/Archives/edgar/data/1604821/000104746915005154/a2224899zs-1.htm | 2015-06-01 | issuer filing; reliability A, full | LVP VIII 6,477,000 shares/warrants/10.4% prior to offering; Mhatre is one of four LUGP directors sharing power; no disclosed board-observer role
• ★★ | Guardant Health S-1 | https://www.sec.gov/Archives/edgar/data/1576280/000162828018011647/guardanthealths-1.htm | 2018-09-05 | issuer filing; reliability A, full | Lightspeed 5,187,072 shares/5.4% before offering; Mhatre one of five GP directors sharing power, not an issuer director
• ★★ | Blend S-1 | https://www.sec.gov/Archives/edgar/data/1855747/000119312521194971/d162671ds1.htm | 2021-06-21 | issuer filing; reliability A, full | Lightspeed IX/Select/Select III 77,181,722 Class A/13.3% before offering; shared GP power; no Mhatre board role
• ★★ | Aurora SPAC S-4 | https://www.sec.gov/Archives/edgar/data/1828108/000119312521216134/d184562ds4.htm | 2021-07-15 | issuer transaction filing; reliability A, full | business-combination and ownership control, not an IPO; implied valuation is not investor proceeds or return
• ★★★ | Forbes Mhatre profile | https://www.forbes.com/profile/ravi-mhatre/ | updated 2026-05-27 | reported profile reviewed; reliability B, full | reported $50b practice, Anthropic lead and exit values; not audited returns or an adviser-filing substitute
• ★★★ | Forbes 2020 firm profile | https://www.forbes.com/sites/alexkonrad/2020/04/09/vc-firm-lightspeed-raises-4-billion/ | 2020-04-09 | reported profile reviewed; reliability B, full | six-person formation, vehicle scale and platform thesis; contemporaneous Light Reading record controls formation scope
• ★★ | Lightspeed Turns $10m into $1b with Nutanix IPO | https://www.forbes.com/sites/alexkonrad/2016/09/30/lightspeed-turns-10-million-into-1-billion-with-nutanix-ipo/ | 2016-09-30 | reported profile; reliability B, partial | reported Lightspeed vehicle-level claim; not personal or audited return
• ★ | Signal investor profile | https://signal.nfx.com/investors/ravi-mhatre | current; reviewed 2026-08-23 | third-party deal database; reliability C, partial | processed for discovery only, then excluded from the 91-card denominator because its 100-row association set has high false-attribution risk
• ★ | SEC Form D appearance index | https://marketcast.smartkarma.com/person/ravi-mhatre | current; reviewed 2026-08-23 | third-party filing index; reliability C, full | processed as a discovery index only; 61 entities/167 appearances were not treated as verified deals without canonical EDGAR support
• ★★★ | SEC IAPD adviser summary | https://adviserinfo.sec.gov/firm/summary/160187 | current; accessed 2026-08-23 | regulator record; reliability A, full | LSVP registration effective 2025-04-28 after ERA withdrawal; firm status, not personal conduct
• ★★★ | Lightspeed Form ADV | https://reports.adviserinfo.sec.gov/reports/ADV/160187/PDF/160187.pdf | other-than-annual amendment filed 2026-05-06; accessed 2026-08-23 | regulator filing, Item 5.F(2)(c); reliability A, full | Lightspeed Management Company, L.L.C. reports adviser-level RAUM of $50,005,412,721 across 51 accounts; adviser perimeter/scale only, never Mhatre AUM, fund value or return
• ★★ | Lightspeed legal and entity disclosures | https://lsvp.com/legal/ | current; accessed 2026-08-23 | official legal page; reliability A/C, full | LSVP, LSIP and Faction boundaries and marketing/disclaimer controls
• ★★ | Stock Slump Not Scaring VC Firms | https://www.sfgate.com/business/article/Stock-Slump-Not-Scaring-VC-Firms-2702112.php | 2000 | contemporaneous press reviewed; reliability B, full | formation and dot-com-market context corroboration, used only alongside the WPG continuity record
• ★★ | The Future of Venture Capital | https://lsvp.com/thefutureofvc/the-future-of-venture-capital/ | current; reviewed 2026-08-23 | official institutional history; reliability A/C, full | interested current narrative compared with the contemporaneous WPG record rather than used to smooth the four/six-founder discrepancy
• ★★★ | Netskope final prospectus | https://www.sec.gov/Archives/edgar/data/2063196/000119312525207534/netskope_424b4_sept_2025.htm | 2025-09 | issuer filing; reliability A, full | Lightspeed 64,493,207 Class B/19.3% pre/16.9% post/19.2% post vote; Arif board; paper value is not proceeds/return
• ★★ | HP acquires Bromium | https://www.hp.com/us-en/newsroom/press-releases/2019/hp-announces-acquisition-of-bromium.html | 2019-09-19 | acquirer announcement; reliability A/C, full | acquisition outcome control against stale Private card status; no investor-proceeds disclosure

Adverse, legal, critical and historical-context routes

• ★★★ | In re Nutanix Stockholder Derivative Litigation dismissal | https://www.govinfo.gov/content/pkg/USCOURTS-cand-3_19-cv-03817/pdf/USCOURTS-cand-3_19-cv-03817-2.pdf | 2020-12-22 | federal-court dismissal stipulation; reliability A, full | Mhatre named in board capacity; voluntarily dismissed, no misconduct finding
• ★★ | Bhonagiri v. Pandey docket | https://dockets.justia.com/docket/california/candce/5%3A2019cv03817/344350 | 2019–2020 | docket index reviewed; reliability B, partial | processed to distinguish derivative action from securities litigation; the federal dismissal record governs disposition and supplies no Mhatre misconduct finding
• ★★ | MuleSoft merger shareholder complaint | https://regmedia.co.uk/2018/04/06/mulesoftcomplaint.pdf | 2018-04 | complaint reviewed; reliability A for allegation only, full | board/vehicle allegations are not findings; issuer supplement and federal docket provide the necessary denial/procedural boundary
• ★★★ | MuleSoft merger litigation supplement | https://www.sec.gov/Archives/edgar/data/1108524/000119312518132739/d552163d424b3.htm | 2018-04-23 | company filing; reliability A for company disclosure, full | litigation allegations and company denial; neither is a misconduct finding
• ★★ | MuleSoft merger complaint docket | https://dockets.justia.com/docket/california/candce/3%3A2018cv02095/324897 | 2018 | federal docket index; reliability B, partial | correct N.D. Cal. proceeding route; preserves allegation-versus-disposition boundary
• ★★ | 2015 Geektime bubble/misses interview | https://www.geektime.co.il/an-interview-with-ravi-mhatre/ | 2015-11-30 | translated direct interview; reliability A/B, full | contemporaneous price/FOMO discipline, claimed one-in-100 selection and desire to inspect misses
• ★ | FTX/broader Lightspeed crypto adverse route | https://medium.com/lightspeed-venture-partners/building-blockchain-infrastructure-one-startup-at-a-time-meet-venture-partner-anand-iyer-a2bdafb25d56 | 2022-07-20 | firm essay reviewed; reliability A, full | processed solely as firm-platform counterevidence; no evidence Mhatre led FTX and no person-level outcome is inferred
• ★ | Broader India portfolio adverse route | https://www.forbesindia.com/article/vc-and-pe-special/lightspeed-the-fund-that-bets-on-getting-in-early/54491/1 | 2019 | reported firm profile reviewed; reliability B, full | processed for India vehicle/team context; OYO, BYJU'S and ShareChat are not person-attributed without deal-specific proof
• ★★ | NVCA/PitchBook Venture Monitor | https://nvca.org/research/pitchbook-nvca-venture-monitor/ | current historical series | industry data; reliability A/B, full | financing, fundraising, exit and valuation context by decision year
• ★★ | SEC IPO prospectus archive | https://www.sec.gov/edgar/search/ | current; searched 2026-08-23 | regulator database; reliability A, full | processed as the discovery route for canonical issuer filings; exact filing URLs, not search results, support ownership and board claims
• ★ | Stanford Libraries / Silicon Valley Archives search | https://searchworks.stanford.edu/ | current; searched 2026-08-23 | catalog/finding-aid route; reliability A, search only | processed with no dedicated Mhatre oral-history or personal-collection hit in the public catalog
• ★ | Computer History Museum oral histories | https://www.computerhistory.org/collections/oralhistories/ | current; searched 2026-08-23 | archive route; reliability A, search only | processed with no dedicated Ravi Mhatre oral-history hit; no inaccessible speech was reconstructed
• ★★★ | BYJU'S insolvency process and orders | https://ibbi.gov.in/claims/order-process/U80903KA2011PTC061427 | current process accessed 2026-08-23 | regulator process record; reliability A, full | CIRP/status control; no Mhatre-specific conduct, stake or loss finding
• ★★★ | Supreme Court BYJU'S judgment | https://api.sci.gov.in/supremecourt/2024/35406/35406_2024_1_1502_56620_Judgement_23-Oct-2024.pdf | 2024-10-23 | court judgment; reliability A, full | procedural/governance outcome control; no finding of misconduct by Mhatre

Scientific correction controls (outside the spoken-appearance denominator)

• ★★★ | NHGRI Human Genome Project FAQ | https://www.genome.gov/Pages/Education/Smithsonian_Exhibition/Human_Genome_Project_FAQ.pdf | current reference; accessed 2026-08-23 | U.S. government scientific reference; reliability A, full | approximately three-billion-base-pair control for Mhatre's inaccurate 2020 aside
• ★★★ | HIV.gov recommended immunizations | https://www.hiv.gov/hiv-basics/staying-in-hiv-care/other-related-health-issues/immunizations-recommended-for-people-living-with-hiv | current reference; accessed 2026-08-23 | U.S. government health reference; reliability A, full | no currently available preventive or therapeutic HIV vaccine; control only, not another appearance

Mirrors, author archive and duplicate assets (excluded from work and appearance denominators)

• ★ | Aurora public-company essay — official migration mirror | https://lsvp.com/stories/driving-innovation-aurora-goes-public/ | current mirror reviewed 2026-08-23 | same underlying 2021 work as the mapped Medium original; reliability A/C, full | migration/canonicalization evidence only; does not create a 41st writing
• ★ | Enabling a Blameless Culture — Medium original | https://medium.com/lightspeed-venture-partners/enabling-a-blameless-culture-ce21e9454287 | 2019-03-20 | migrated original reviewed 2026-08-23; reliability A/C, full | structured metadata is incomplete; official Lightspeed body signature controls Ravi Mhatre/Jay Madheswaran coauthorship; one underlying work
• ★ | Ravi Mhatre Medium author archive | https://ravi-lsvp.medium.com/ | accessed 2026-08-23 | intermittent author-index route; reliability C, partial | processed as a discovery/denominator boundary only; fragile pagination prevents an exhaustive lifetime-work claim
• ★ | Hill & Valley Forum Apple podcast mirror | https://podcasts.apple.com/us/podcast/the-hill-valley-forum-podcast/id1692653857 | reviewed 2026-08-23 | same underlying 2025 forum appearance as the mapped captioned YouTube family; reliability B/C, metadata | mirror only; does not create a 32nd appearance
• ★ | Bloomberg 2026 DailyMotion mirror | https://www.dailymotion.com/video/xa0qjce | reviewed 2026-08-23 | same underlying Bloomberg appearance as the mapped captioned YouTube family; reliability B/C, full video | mirror only; does not create a 32nd appearance
• ★ | TiEcon 2012 excerpt asset | https://www.youtube.com/watch?v=BPcxfbezsvA | reviewed 2026-08-23 | excerpt of the mapped S79P83FMZ0U appearance; reliability B/C, captioned | duplicate/excerpt only; does not create a 25th YouTube family
• ★ | MuleSoft 2018 longer event asset | https://www.youtube.com/watch?v=LM7kZsSRNBM | reviewed 2026-08-23 | same underlying appearance family as mapped excerpt _AO2_Uzpdz4; reliability B/C, captioned | complementary/longer asset within one family, not an additional appearance
• ★ | NYSE 2024 duplicate upload | https://www.youtube.com/watch?v=i85E4mJuWYE | reviewed 2026-08-23 | same-channel duplicate of canonical X90SGz1BivU; reliability B/C, no captions | duplicate only; one OpenAI-transcribed family, not an additional appearance

Official-profile company-card denominator for DEALS

The live Ravi Mhatre profile exposed the following 91 unique company-card anchors in this exact order at approximately 2026-08-23T13:09Z. Each card proves only a current official-profile association; it does not by itself prove Mhatre led or joined an investment, sat on the board, selected a vehicle, controlled capital, owned securities, or received proceeds.

• ★★ | 01 · Company card — Abstract | https://lsvp.com/company/abstract/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 02 · Company card — AppDynamics | https://lsvp.com/company/appdynamics/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 03 · Company card — Alchemy | https://lsvp.com/company/alchemy/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 04 · Company card — Aqua Security | https://lsvp.com/company/aqua-security/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 05 · Company card — Aurora | https://lsvp.com/company/aurora/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 06 · Company card — Blameless | https://lsvp.com/company/blameless/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 07 · Company card — Blend | https://lsvp.com/company/blend/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 08 · Company card — Bromium | https://lsvp.com/company/bromium/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 09 · Company card — BYJU'S | https://lsvp.com/company/byjus/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 10 · Company card — Cheddar | https://lsvp.com/company/cheddar/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 11 · Company card — Clever | https://lsvp.com/company/clever/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 12 · Company card — Confluera | https://lsvp.com/company/confluera/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 13 · Company card — Descope | https://lsvp.com/company/descope/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 14 · Company card — Dexterity | https://lsvp.com/company/dexterity/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 15 · Company card — EdgeSpring | https://lsvp.com/company/edgespring/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 16 · Company card — Exabeam | https://lsvp.com/company/exabeam/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 17 · Company card — Fiddler | https://lsvp.com/company/fiddler/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 18 · Company card — Glean | https://lsvp.com/company/glean/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 19 · Company card — Guardant Health | https://lsvp.com/company/guardant-health/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 20 · Company card — Influitive | https://lsvp.com/company/influitive/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 21 · Company card — Karius | https://lsvp.com/company/karius/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 22 · Company card — Kodiak AI | https://lsvp.com/company/kodiak/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 23 · Company card — Kosmix | https://lsvp.com/company/kosmix/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 24 · Company card — Magic Eden | https://lsvp.com/company/magic-eden/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 25 · Company card — Masergy | https://lsvp.com/company/masergy/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 26 · Company card — Materialize | https://lsvp.com/company/materialize/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 27 · Company card — Multiverse | https://lsvp.com/company/multiverse/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 28 · Company card — MuleSoft | https://lsvp.com/company/mulesoft/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 29 · Company card — Mysten Labs | https://lsvp.com/company/mysten-labs/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 30 · Company card — Offchain Labs | https://lsvp.com/company/offchain-labs/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 31 · Company card — Osmos | https://lsvp.com/company/osmos/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 32 · Company card — Natera | https://lsvp.com/company/natera/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 33 · Company card — Nutanix | https://lsvp.com/company/nutanix/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 34 · Company card — Orbis Education | https://lsvp.com/company/orbis-education/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 35 · Company card — Pulse | https://lsvp.com/company/pulse/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 36 · Company card — Qubole | https://lsvp.com/company/qubole/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 37 · Company card — RAPsphere | https://lsvp.com/company/rapsphere/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 38 · Company card — Reflektive | https://lsvp.com/company/reflektive/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 39 · Company card — Riverbed Technology | https://lsvp.com/company/riverbed-technology/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 40 · Company card — Roofstock | https://lsvp.com/company/roofstock/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 41 · Company card — Rubrik | https://lsvp.com/company/rubrik/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 42 · Company card — SailPoint | https://lsvp.com/company/sailpoint/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 43 · Company card — Salto | https://lsvp.com/company/salto/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 44 · Company card — Sentio | https://lsvp.com/company/sentio/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 45 · Company card — Snorkel AI | https://lsvp.com/company/snorkel-ai/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 46 · Company card — StreamOnce | https://lsvp.com/company/streamonce/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 47 · Company card — Talon | https://lsvp.com/company/talon/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 48 · Company card — Tonkean | https://lsvp.com/company/tonkean/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 49 · Company card — TutorVista | https://lsvp.com/company/tutorvista/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 50 · Company card — ThoughtSpot | https://lsvp.com/company/thoughtspot/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 51 · Company card — TheFind | https://lsvp.com/company/thefind/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 52 · Company card — Typeface | https://lsvp.com/company/typeface/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 53 · Company card — Udaan | https://lsvp.com/company/udaan/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 54 · Company card — Virsa Systems | https://lsvp.com/company/virsa-systems/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 55 · Company card — Waveset | https://lsvp.com/company/waveset/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 56 · Company card — Yellow.ai | https://lsvp.com/company/yellow-ai/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 57 · Company card — Yugabyte | https://lsvp.com/company/yugabyte/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 58 · Company card — Zscaler | https://lsvp.com/company/zscaler/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 59 · Company card — Anduril | https://lsvp.com/company/anduril/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 60 · Company card — Mithril | https://lsvp.com/company/mithril/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 61 · Company card — Augment | https://lsvp.com/company/augment/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 62 · Company card — Nexus | https://lsvp.com/company/nexus/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 63 · Company card — Skild AI | https://lsvp.com/company/skild-ai/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 64 · Company card — Helsing | https://lsvp.com/company/helsing/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 65 · Company card — Clutch | https://lsvp.com/company/clutch/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 66 · Company card — Pacific Fusion | https://lsvp.com/company/pacific-fusion/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 67 · Company card — Cartesia | https://lsvp.com/company/cartesia/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 68 · Company card — Eve | https://lsvp.com/company/eve/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 69 · Company card — Anthropic | https://lsvp.com/company/anthropic/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 70 · Company card — Neuralink | https://lsvp.com/company/neuralink/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 71 · Company card — Antares Therapeutics | https://lsvp.com/company/antares-therapeutics/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 72 · Company card — Nominal | https://lsvp.com/company/nominal/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 73 · Company card — Thinking Machines | https://lsvp.com/company/thinking-machines/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 74 · Company card — Arena | https://lsvp.com/company/arena/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 75 · Company card — Odyssey Therapeutics | https://lsvp.com/company/odyssey-therapeutics/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 76 · Company card — Zag Bio | https://lsvp.com/company/zag-bio/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 77 · Company card — Rain | https://lsvp.com/company/rain/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 78 · Company card — Character.AI | https://lsvp.com/company/character-ai/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 79 · Company card — Netskope | https://lsvp.com/company/netskope/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 80 · Company card — Periodic Labs | https://lsvp.com/company/periodic-labs/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 81 · Company card — Phantom | https://lsvp.com/company/phantom/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 82 · Company card — Unconventional AI | https://lsvp.com/company/unconventional-ai/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 83 · Company card — Raindrop | https://lsvp.com/company/raindrop/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 84 · Company card — Castelion | https://lsvp.com/company/castelion/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 85 · Company card — Ricursive Intelligence | https://lsvp.com/company/ricursive-intelligence/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 86 · Company card — Temporal | https://lsvp.com/company/temporal/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 87 · Company card — Nexthop AI | https://lsvp.com/company/nexthop-ai/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 88 · Company card — Fireworks AI | https://lsvp.com/company/fireworks-ai/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 89 · Company card — Ineffable Intelligence | https://lsvp.com/company/ineffable-intelligence/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 90 · Company card — Science Corporation | https://lsvp.com/company/science-corporation/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence
• ★★ | 91 · Company card — Helion | https://lsvp.com/company/helion/ | current; observed 2026-08-23T13:09Z | official profile-linked company card; reliability A/C, full | association only; role, stage, vehicle, board, ownership, economics and outcome require separate evidence

Completed-peer comparison controls

• ★★★ | David Cowan — Road Map Investing | https://whohastimeforthis.blogspot.com/2005/08/road-map-investing.html | 2005-08 | original investor writing; reliability B, full | market-map, sourcing, stale-map and failure-process comparator; not a return comparison
• ★★★ | Danny Rimer — Newcomer interview/transcript | https://podcast.newcomer.co/episode/index-ventures-danny-rimer-talks-figmas-ipo-and-vc-bets | 2025 | edited direct-voice transcript; reliability B, full | global platform, founder service, cycles and current AI comparator; not an alpha comparison
• ★★★ | Arthur Patterson and Jim Swartz — 20VC | https://thetwentyminutevc.libsyn.com/20vc-special-accel-founders-arthur-patterson-and-jim-swartz-on-building-accel-into-one-of-the-most-prominent-venture-firms-over-four-decades-how-todays-market-compares-to-the-dot-com-bubble-how-to-do-generational-transition-well-and-why-accel-will-never-h | 2022 | edited direct-voice audio/readout; reliability B, full | prepared mind, accountable partner, cycles, vehicles and succession comparator
• ★★★ | David Cowan / Bessemer — Rocket Lab 13D | https://www.sec.gov/Archives/edgar/data/1819994/000119312521266638/d221791dsc13d.htm | 2021-09-01 | regulator filing; reliability A, full | peer vehicle/entity control; no cross-investor return inference
• ★★★ | Danny Rimer / Index — Figma S-1 | https://www.sec.gov/Archives/edgar/data/1579878/000162828025033742/figma-sx1.htm | 2025 | issuer filing; reliability A, full | peer board/holder/company control; no personal economics or cross-investor return inference

Coverage assessment

Search scope and variants

The discovery sweep used Ravi Mhatre, Ravi Brian Mhatre, Ravi LSVP, Mhatre Lightspeed, spelling collision Ravi Mahtre, and combinations with interview, transcript, podcast, video, essay, portfolio, board, S-1, 13G, Form 4, ADV, returns, criticism, underperformance, lawsuit, complaint, SEC, FTX, misses and decade/year terms. Venue sweeps covered Lightspeed/Medium, Stanford eCorner/STVP, TechCrunch/StrictlyVC, Bloomberg, Jefferies, Newcomer, YouTube, Milken, Inc42, Geektime, Economic Times, Forbes/Forbes India, SEC EDGAR/IAPD, federal court indexes, Stanford SearchWorks and the Computer History Museum.

Category and access accounting

The map contains 227 checked checklist rows: 40 writings, 31 spoken appearances, 36 firm/company/filing routes, 14 adverse/context routes, 2 scientific correction controls, 5 completed-peer controls, 8 mirror/archive/duplicate-asset routes and the separate 91-card association denominator. The 136 research and provenance rows excluding the card denominator remain analytically distinct from those 91 association rows. Across the full map there are 231 explicit URL occurrences and 229 unique URLs because some appearance families preserve complementary or canonical-plus-edited routes. Geektime is deliberately cross-listed as both direct voice and adverse evidence, while the Anand Iyer essay is deliberately cross-listed as a writing and a broader-firm crypto boundary; neither cross-listing counts as an independent underlying source. All 16 rows that entered REVIEW unchecked were individually re-opened or their named search route re-run and are now recorded as used, discovery-only, allegation-only, or no-hit/barrier—not silently treated as positive evidence.

Direct voice and archive routes

The Stanford 2020 transcript is the best complete direct-voice source; its four clip transcripts are derivative representations of the same talk and count once. The 2017 TechCrunch/StrictlyVC edited Q&A is joint with Barry Eggers. Jefferies and ET are edited; Inc42 is moderator notes; Geektime is a Hebrew rendering; Newcomer exposes full audio/video but no publisher transcript. Provider captions supported 21 YouTube families. AppDynamics, Bloomberg Crypto and NYSE exposed neither provider captions nor an official transcript, so their legitimately acquired audio was transcribed one at a time on 2026-08-23 with OpenAI whisper-1 verbose segment timestamps; the machine text was reviewed for propositions, but neither full transcripts nor media are committed. Bloomberg Nutanix 2016 exposes only metadata and remains the one partial family. No public LP letters, IC memos, board books, shareholder letters by Mhatre, dedicated oral history, personal archive or complete public talk catalog was found. SearchWorks and CHM were attempted without a dedicated collection hit.

Candidate portfolio denominator and attribution controls

The current profile's narrative highlights 10 companies—Anduril, AppDynamics, Guardant, MuleSoft, Natera, Nutanix, Offchain Labs, Rubrik, ThoughtSpot and Zscaler—while its live card layer exposed 91 unique company anchors in exact order at approximately 2026-08-23T13:09Z. The ten are subsets of the 91; every card is association evidence only. The ledger deduplicates follow-ons inside each company family and separates Mhatre-led or board-attributed evidence; shared Lightspeed vehicles; another named partner; LSVP firm-wide; legally distinct LSIP; personal/family entities; and role-only/false-positive paths. Third-party databases remain discovery-only because they conflate those categories. Shared voting or dispositive power in a filing is not personal ownership, cost basis, proceeds or return.

Historical context and completed-peer routes

NVCA/PitchBook annual data and contemporaneous issuer/press records can frame the dot-com reset, 2007–2009 enterprise financing, 2015–2016 IPO market, 2020–2021 capital abundance, 2022 reset and 2025–2026 AI concentration. Closest completed comparisons are David Cowan for enterprise system maps and staged evidence, Arthur Patterson for firm formation/prepared-market work, and Danny Rimer for global LP-backed platform scale. Comparison must match dimensions rather than infer relative returns.

Highest-value retrieval leads

1. Seek an archival transcript, caption track or legitimate audio route for Bloomberg's 2016 Nutanix interview, and recheck the three OpenAI-derived notes if an official transcript later appears.
2. Reconcile the selected Medium corpus against current Lightspeed migrations and collapse mirrors without losing underlying-work or coauthor accounting.
3. Reconcile official portfolio image metadata, filing-derived historical boards and direct-essay deals into a stable candidate union before computing any count.
4. Trace the MuleSoft complaint to its official docket and disposition; keep the Nutanix voluntary dismissal distinct from allegations.
5. Seek a separately accessible adviser brochure, later Form ADV amendments and fund/entity records beyond the already mapped current Form ADV; private carry, LP cash flows, reserves and IC allocation remain unavailable.
6. Seek contemporaneous Series A announcements and founder accounts for AppDynamics, MuleSoft, Nutanix and Rubrik to reduce retrospective-deal bias.

Material gaps after reasonable discovery

No complete opportunity/pass/write-off ledger, audited personal or vehicle cash-flow history, security-by-security cost/dilution/distribution chain, carry allocation, IC minutes, board files, LP communications, loss denominator, opportunity funnel or partner-work allocation is public. Economic contribution and Mhatre-specific alpha therefore cannot be computed from this source map. The research record can support decision-process and attribution analysis more strongly than a personal-performance claim.

Source notes