Y Combinator

Confidence 0.95 · 14 sources · last confirmed 2026-06-23

Y Combinator is a Silicon Valley startup accelerator founded in 2005. President & CEO: Garry Tan (since 2023). Operates the Startup School educational programme + biannual accelerator batches + Bookface internal social platform.

Promoted to a wiki entity page on 12 May 2026 after appearing as author: (channel) on two ingested videos in two consecutive days (Tan 23 April + Hu 24 April). Bumped to source_count: 3 on 14 May 2026 with the addition of Replit on Founder Firesides, landing one day after Hu and forming a three-day YC-channel cluster on the AI-native-company thesis from three distinct vantages (President / Partner / Portfolio-founder). Bumped to source_count: 4 on 21 May 2026 with the addition of AnswerThis on YC Root Access (19 May 2026), extending the anchor triple into a quad and adding the 2-FTE-startup-scale portfolio-founder vantage on a fourth YC channel (YC Root Access). Bumped to source_count: 5 on 21 May 2026 (same day as Garg) with the addition of Campfire on YC Root Access — the enterprise-B2B founder-vantage worked example of the AI-native-company thesis: tech-company-targeted AI-native ERP pulling enterprise customers off NetSuite from Q4 2024 onward, ARR more-than-doubling each quarter since. With Glasgow, YC Root Access becomes the second YC channel (after the Garg ingest) to land two ingested sources, and is promoted to alias on this entity page so the dangling-authors lint resolves it back to Y Combinator (same convention as Anthropic’s Claude channel alias). Bumped to source_count: 10 on 22 May 2026 with the addition of three earlier YC Root Access Founder Firesides episodes that backfill the channel-cluster from February–April 2026: Letter AI (25 Feb), Momentic (23 March), and Luminai (9 April). With these three ingests YC Root Access becomes the YC channel with the deepest wiki coverage (six sources spanning Feb–May 2026), and the Founder Firesides format gains three vertical-AI portfolio-vendor vantages — sales-enablement (Letter AI), QA/testing (Momentic), and healthcare-administrative automation (Luminai).

Role in the wiki

YC is the AI-native-startup-formation vantage the wiki tracks across ten distinct 2026-batch-contemporaneous sources spanning February–May 2026 and four distinct YC channels (Tan’s personal channel / Startup School / Founder Firesides / YC Root Access). The 22 May 2026 batch ingest backfilled the channel-cluster with three earlier Founder Firesides episodes on YC Root Access (Letter AI / Momentic / Luminai — sections 7–9 below) that pre-date the spring 2026 partner-content burst and re-anchor the cluster on the portfolio-founder vantages of vertical-AI Series A/B graduates:

1. Tan / GStack (Tan 2026)

YC President & CEO Garry Tan publicly shipping GStack — an open-source toolkit of skills wrapping Claude Code with named primitives (Office Hours / Adversarial Review / Design Shotgun / Code Review / SLQA / Ship). 3-week build velocity; “more GitHub stars than Ruby on Rails” by ingest time. The Office Hours skill is “the distilled version of what is thousands and tens of thousands of hours that the 16 YC partners have spent many many years honing and perfecting” — a 10% strength version of YC’s real partner-mediated adversarial review of startup ideas.

2. Hu / Startup School (Hu 2026)

YC Partner Diana Hu (published 24 April 2026, the day after Tan) presents the prescriptive framework for AI-native company structure: AI-as-operating-system; closed-loop vs open-loop companies; software factories; the 1,000× engineer thesis; three archetypes (IC / DRRI / AI founder type) via Jack Dorsey at Block; token-maxing not headcount-maxing.

3. Masad / Founder Firesides (Masad 2026)

Replit CEO Amjad Masad in conversation with YC’s Andrew Miklas (published 25 April 2026, the day after Hu). Masad gives the portfolio-founder-vantage worked example of the YC-batch thesis: the company-of-the-future is built of builders + salespeople; Replit’s internal vibe-coding-resident team (a generalist team with vague mission “go around the company and make it better”) is the closed-loop company pattern in operation; the skills-call revolution at the Replit-platform layer parallels Tan’s GStack-style packaging.

4. Garg / YC Root Access (Garg 2026)

AnswerThis founder Ayush Garg on YC Root Access (published 19 May 2026), one slot in a multi-talk YC session on agent-harness engineering — Garg references “Pete and Tom and Gary” as prior speakers in the same session (likely Pete Koomen, Tom Blomfield, Garry Tan). Garg presents the 2-FTE-startup-scale worked example of the YC-batch thesis: $2M ARR with two full-time employees + 2–3 contractors, running on an internal AI ops agent (Claude Code CLI wrapped in Python; self-extending coding sub-agent that has authored 45+ CLIs; agent-editable instructions.md ratcheted by Slack feedback). Names the three-memory framework (factual / behavioural / procedural) the wiki has been carrying without an anchor. Closes the talk with an explicit recipe-to-clone: “if you want to copy this entire agent, just take a picture of this and give it to your cloud code.”

5. Glasgow / YC Root Access (Glasgow 2026)

Campfire founder/CEO John Glasgow (YC S23) on YC Root Access (published 20 May 2026, one day after Garg). The enterprise-B2B founder-vantage worked example of the AI-native-company thesis — a fireside on what AI-native looks like in the most ancient enterprise-software category in tech (NetSuite-class ERP). The substantive contribution is the “AI-native safety inversion” observation: from Q4 2024 onward, the buyer-side narrative on enterprise software procurement flipped — being the incumbent meant you were not AI-native, and boards explicitly authorised buying AI-native vendors even when day-to-day operators were not yet ready. Companion claims: tech-stack-turnover thesis for the why now (finance stack already modernised — payroll, spend management — except the core general ledger); wedge displacement over feature-completeness (tech-company customers used ~10% of NetSuite; Campfire owned that slice and was pulling customers off NetSuite within nine months); founder-led-sales-to-$1M-ARR doctrine even in the AI era (“I do really recommend founders to stay in the founder sales mode… offloading it to AI, offloading it to some AE… feels like, oh let’s just bring in a professional”); product velocity as enterprise-buyer trust signal at four-person-company scale; own-foundation-model + custom-agent-platform as competitive moat in the AI-native ERP cohort.

6. Tan & Hu / Stanford CS153 (Tan & Hu 2026)

YC President & CEO Garry Tan and YC GP Diana Hu guest-lecturing in Stanford’s CS153 Frontier Systems course (published 20 May 2026, same day as Glasgow on YC Root Access — the YC-channel cluster now extends to two-per-day publishing cadence). The lecture is the fifth-and-sixth-anchor-bundled-into-one for the wiki’s YC corpus: same-day publication as Glasgow but at a substantially different altitude — Stanford classroom, two YC operating-leadership voices in dialogue. The course host explicitly frames the lecture as the company-formation-layer counterpart to earlier CS153 lectures on compute (Jensen Huang), capital (Ben Horowitz), and energy/chips (Scott Nolan), positioning YC’s introduction of the SAFE in the 2010s as the standardisation moment for venture capital comparable to electrical-grid buildout, and arguing that what’s happening now at the cognitive layer is a parallel standardisation — “this time it’s code, and not just code, markdown is code, literally the new ___.” The substantive contribution is the agentic-primitives-map-to-company-structure framing — skills = employees, resolvers = org chart, GBrain = internal process, check-resolvable = audit and compliance, trigger evals = performance reviews — the wiki’s first founder-CEO-altitude rhetorical claim that the agent-harness primitives are structurally identical to a company’s org structure, not merely metaphorical. Companion claims: the 5-day Posterous rebuild on $200/month Claude Max plan (10 people / $4M / 2 years collapsed to 1 founder / 5 days / $200); the closed-loop-vs-open-loop control-systems framing (Hu’s contribution, made explicit with P-controller analogy); the $1–2M revenue-per-employee benchmark re-anchored to Salient (voice agents for loan servicing), Happy Robot (freight forwarders; Series B; 10× revenue in a year), Reducto (document processing infrastructure); the cross-modal evals as a Skillify built-in (Opus 4.6 + GPT-5.5 + Deepseek V4 cross-evaluating); the GBrain extension to a typed knowledge graph + in-progress epistemology layer for “hunches vs beliefs vs world knowledge.” Hu closes with the Anthropic economic-deployment chart and the “giant white space across back office, finance, data, academics, cybersecurity, customer service — room for hundreds of AI unicorns” call-to-action.

7. Akhtar & Forget / Letter AI (Akhtar & Forget 2026)

Letter AI CEO Ali Akhtar and CTO Armen Forget in conversation with Diana Hu on YC Root Access Founder Firesides (published 25 February 2026; the earliest YC Root Access ingest in the wiki). $40M Series B announcement. The pivot-during-batch + Lenovo-closed-in-batch anchor pair: Letter AI was originally Tractatus (developer tools for generative AI); pivoted in-batch to AI-native sales enablement; closed Lenovo as a customer during the batch via a former colleague’s warm intro. Wiki’s clearest founder-vantage worked example of MCP-server + agent-to-agent protocol as B2B product surface for a vertical-SaaS vendor — “salespeople now are sometimes in Cursor and they’re doing their own research, and now they can use a Letter AI MCP server to get the content.”

8. Wu & An / Momentic (Wu & An 2026)

Momentic co-founders Weiwei Wu and Jeff An (YC W24) in conversation with Harj Taggar on YC Root Access Founder Firesides (published 23 March 2026). $15M Series A announcement. The verification-layer-for-software / truth-driven-development anchor — Momentic does functional testing (impersonates end users) and integrates with Cursor and Claude Code via MCP as the closing-the-loop-for-coding-agents product surface. Notion runs ~500K test runs/day on Momentic with “Momentic tests must pass before one of Notion’s engineers can merge their PR.” Productive contradiction filed with the Tan & Hu CS153 1,000×-engineer thesis: Wu’s “Codex only makes you a 10x engineer if you weren’t a 10x engineer to begin with” is a direct rhetorical inversion — the multiplier as conditional on existing skill rather than unconditional.

9. Dinakaran / Luminai (Dinakaran 2026)

Luminai founder/CEO Kesava Kirupa Dinakaran (YC S20) in conversation with Aaron Epstein on YC Root Access Founder Firesides (published 9 April 2026). $38M Series B (~$60M lifetime). The paper-and-fax-vertical-AI-wedge anchor at trillion-dollar-industry scale — the Cleveland Clinic worked example (“the way that most patients get referred into the Cleveland Clinic is through a fax”) + the verticalisation-decision template (horizontal automation platform → all-in on healthcare → narrow further to a specific set of wedge use-cases) + the enterprise-sales-as-relational-game doctrine + the LinkedIn-contact-scraping script. Companion to Vori (grocery) as the wiki’s second trillion-dollar-paper-and-fax-industry exemplar. Mid-interview tangent connects Dinakaran’s Rubik’s-cube-as-childhood-skill to Andrej Karpathy’s YouTube tutorials.

10. Warren / Startup School (Warren 2026)

YC Visiting Partner Charlie Warren on Startup School (published 3 June 2026, the second Startup School ingest after Hu / April 2026). The wiki’s first YC source explicitly framing AI-native services companies as a distinct go-to-market category — alongside the existing AI-software-product cluster (Hu / Tan / Masad / etc.), Warren writes the services-company sibling of Hu’s playbook: same YC altitude, same built-from-scratch on AI core thesis, different category (services that sell outcomes, not software products). Markets named as YC-internal known-good-fit: tax / audit / insurance / mortgages / parts of healthcare / parts of logistics. The four market-selection traits Warren introduces (low trust + low task-level judgment + high intelligence threshold + regulation-as-moat); the Sam Altman Test for model-disruption-resistance; the three founder attributes (domain fluency + model fluency + operational rigor); the outcome-as-product vs co-pilot-as-product distinction as the central category-defining frame. Two YC-portfolio worked examples named: Panacea (FDA regulatory services for biotech/medtech — regulation-as-moat + outcome-based pricing exemplar) and The General Legal Team (AI-native law firm with Cooley + Fenwick + Casetext founder-team composition). The P&L opportunity-size frame: services 30%-margin floor → AI-leveraged 50%+ on TAMs 2-3× bigger than software. The don’t-buy-your-way-in rule: only valid reason to acquire an existing services business is fast regulatory moat (e.g. insurance licensing). With Warren the YC corpus now spans two AI-native-company vertical curricula — Hu on AI-software products from the ground up, Warren on AI-services companies from the ground up.

11. Kolysh / Startup School — early-stage GTM tactics (Kolysh 2026)

YC Visiting Partner Max Kolysh on Startup School (published 22 June 2026). Distinct from the AI-native-company cluster above — this is a tactical sales playbook for the first 10 customers, compiled from a Bookface founder survey: customers 1–3 from the warm network, 4–10 from doing-things-that-don’t-scale (show up in person, founder dinners, Reddit DMs, advice-framed outreach), and tools (Apollo/Clay/sequences) only mattering at 10–20. Its through-line — “the first 10 customers will come from you, not a tool” — is the GTM-tactics generalisation of the founder-led-sales doctrine the wiki already carries from Luminai and Campfire. Only lightly an AI source (Happenstance network-search, Clay enrichment, and the “sounds-like-an-LLM” cold-email negative signal).

The YC-batch-context anchor — now ten Startup-School-and-Founder-Firesides sources across four channels (eleven counting Koomen / Lightcone)

The ten sources together form a YC-batch-context corpus spanning February–May 2026 and four distinct YC channels:

Vantage	Source	Channel	Contribution
President / platform-toolkit	Tan / GStack	Tan’s personal channel	The packaged-toolkit-ship-at-venture-pace worked example
Partner / prescription	Hu / Startup School	Startup School	The framework — closed-loop companies + software factories + 1,000× engineer
Portfolio-founder / platform-vendor	Masad / Founder Firesides	Founder Firesides	The vendor-CEO worked example — Replit as the floor-raising infrastructure + post-prompting trajectory
Portfolio-founder / 2-FTE-startup	Garg / AnswerThis	YC Root Access	The micro-scale worked example — $2M ARR / 2 FTE on a self-extending Claude Code harness; three-memory framework (factual / behavioural / procedural)
Portfolio-founder / enterprise-B2B vendor	Glasgow / Campfire	YC Root Access	The enterprise-displacement worked example — AI-native ERP pulling >100 customers off NetSuite from Q4 2024 onward; the AI-native safety inversion as the buyer-side procurement-room mechanism
Co-founder pair / partners / Stanford classroom	Tan & Hu / Stanford CS153	Stanford CS153 lecture	The agentic-primitives-map-to-company-structure framing + 5-day Posterous rebuild + cross-modal evals + GBrain epistemology layer
Portfolio-founder pair / sales-enablement vertical-SaaS	Akhtar & Forget / Letter AI	YC Root Access	The MCP-server-as-vendor-product-surface worked example + pivot-during-batch from developer tools to sales enablement + Lenovo-closed-in-batch warm-intro template
Portfolio-founder pair / QA vertical-SaaS	Wu & An / Momentic	YC Root Access	The verification-layer-for-software / truth-driven-development thesis + MCP-callable-external-verifier product surface + Notion 500K-test-runs/day anchor + 10×-counter to the 1,000× thesis
Portfolio-founder / healthcare-administrative vertical-SaaS	Dinakaran / Luminai	YC Root Access	The paper-and-fax-vertical-AI-wedge anchor at trillion-dollar-industry scale (Cleveland Clinic worked example) + the horizontal-to-vertical specialisation decision + the enterprise-sales-as-relational-game doctrine + LinkedIn-contact-scraping warm-intro script
Visiting Partner / AI-native services category	Warren / Startup School	Startup School	The category-defining playbook — outcome-as-product vs co-pilot-as-product; four market-selection traits (low trust + low task-judgment + high intelligence threshold + regulation-as-moat); the Sam Altman Test; three founder attributes (domain + model + operational fluency); Panacea + The General Legal Team worked examples; 30%-services-margin floor → AI-leveraged 50%+ on TAMs 2-3× bigger than software; don’t-buy-your-way-in rule

All ten align on the same diagnosis: AI-native companies need fewer humans + queryable artifacts + software factories + a packaged harness of skills + a buyer-side procurement-narrative flip that grants them air cover against incumbents. The three 22-May backfill ingests (Letter AI / Momentic / Luminai) extend the diagnosis backwards in time and across three vertical-SaaS categories (sales-enablement / QA / healthcare-administrative), each instantiating the MCP-and-agent-protocol-as-vendor-product-surface pattern at the customer-vantage layer.

The Garg and Glasgow ingests in the same week (19 + 20 May 2026) make YC Root Access a major second YC channel (after Startup School) for the wiki’s AI-native-company corpus. The 22 May 2026 batch ingest of three earlier Founder Firesides episodes elevates YC Root Access to the YC channel with the deepest wiki coverage (six sources across Feb–May 2026). Garg’s references to “Pete and Tom and Gary” still suggest companion slots in the same multi-talk agent-harness-engineering session that may be ingestible.

People at Y Combinator (referenced across the wiki)

• Garry Tan — President & CEO since 2023; engineer (Stanford CS); ex-Palantir #10; Posterous co-founder; built first version of Bookface. The AI founder type archetype operationalised at YC-president scale — runs 10-15 parallel Claude Code sessions, ~400 PRs in review. Second-source mention: referenced by name in Garg 2026 (“as Pete and Tom and Gary mentioned”) as a prior speaker in the same YC Root Access multi-talk session. Promote when a second standalone-source mention lands.
• Diana Hu — YC Partner; speaker at YC Startup School on AI-native company construction. Currently Dangling; promote on second-source mention.
• Andrew Miklas — YC, interviewer on Founder Firesides (Masad / 2026-04-25). Currently Dangling; promote on second-source mention.
• Pete Koomen — likely YC partner; referenced in Garg 2026 as a prior speaker in the same YC Root Access session covering the thin harness point. Currently Dangling; promote when his slot in the session is ingested or a second standalone source surfaces.
• Tom Blomfield — likely YC managing director (ex-Monzo / GoCardless founder); referenced in Garg 2026 as a prior speaker in the same YC Root Access session covering the thin harness point. Currently Dangling; promote when his slot in the session is ingested or a second standalone source surfaces.
• Aaron Epstein — YC General Partner; host of the YC Root Access Founder Firesides episode with Dinakaran (Dinakaran 2026). Conducted Dinakaran’s original 2020 YC interview (per the fireside dialogue). Currently Dangling; promote on second-source mention.
• Harj Taggar — YC Managing Partner; host of the YC Root Access Founder Firesides episode with Wu & An / Momentic (Wu & An 2026). Currently Dangling; promote on second-source mention.
• Dalton Caldwell — Managing Director / Group Partner at YC for 10+ years (21 batches as of April 2024). The wiki’s canonical voice on the evergreen YC-partner-wisdom layer — tar pit ideas + pivot template + find-an-incumbent-with-low-NPS + customer-validation-first + the 2024 request-for-startups inventory. Promoted from Dangling to entity page on 22 May 2026 after the Lenny’s Podcast (April 2024) solo-headlining ingest, his second substantive source mention (first was Campfire (May 2026) where he was named as Campfire’s YC partner-of-record and the author of YC’s ERPs as a request-for-startups prompt that Campfire emerged in response to).

People at YC portfolio companies (referenced across the wiki)

• Ayush Garg — founder of AnswerThis; speaker on YC Root Access (Garg 2026). Currently Dangling; promote on second-source mention.
• Ryan — non-technical co-founder at AnswerThis (surname not stated in source); the Slack-feedback-loop-with-the-agent exemplar. Currently Dangling; promote on second-source mention.
• Mukund Jha — founder/CEO of Emergent; second YC-portfolio founder vantage on Claude Code as substrate (Jha 2026). Currently Dangling; promote on second-source mention.
• Kesava Kirupa Dinakaran — founder/CEO of Luminai (YC S20); the paper-and-fax-vertical-AI-wedge anchor for healthcare-administrative automation (Dinakaran 2026). Childhood Rubik’s-cube champion who learned the cube from Andrej Karpathy’s YouTube tutorials. Currently Dangling; promote on second-source mention.
• Ali Akhtar — co-founder/CEO of Letter AI (originally Tractatus, ~W23/S23 batch); the MCP-server-as-vendor-product-surface anchor for sales enablement (Akhtar & Forget 2026). Ex-Director of Engineering for ML at Samsara and project44. Currently Dangling; promote on second-source mention.
• Armen Forget — co-founder/CTO of Letter AI; same ingest. Currently Dangling; promote on second-source mention.
• Weiwei Wu — co-founder of Momentic (YC W24); the truth-driven-development / closing-the-loop-for-coding-agents anchor for the QA vertical (Wu & An 2026). Ex-Robinhood (testing-team experience anchor; saw team grow 300 → 1,000 engineers). Currently Dangling; promote on second-source mention.
• Jeff An — co-founder of Momentic; same ingest. Ex-Cambridge chemistry. Currently Dangling; promote on second-source mention.

Operational practices observed across both ingests

• YC partners as adversarial-review-as-a-service — Hu’s queryable organization framing operates on the same partnerships YC has long offered to portfolio companies; Tan’s Office Hours skill is “the 10% strength version of what we do at YC every day.” Both sources reference YC partner sessions as the human-mediated baseline.
• AI-as-default-tooling-recommendation — both Tan and Hu prescribe Claude Code (or Codex or Cursor) as the agent-runtime; software factories as the architectural pattern; token-maxing as the resource-allocation discipline.
• Open-source-toolkit-ship-at-venture-pace — Tan’s GStack ships in 3 weeks to more-stars-than-Ruby-on-Rails. YC’s institutional preference for shipping-and-iterating is operationalised at the accelerator-president scale.
• The accelerator-batch-context as a wiki-relevant temporal anchor — both sources land in late April 2026, mid-batch for YC’s spring cohort. The accelerator’s partner-public-content cadence now produces wiki-load-bearing sources at scale.

Notable products / programmes

• Startup School — YC’s free online programme + content channel; Hu 2026 published there.
• Founder Firesides — YC’s interview series with portfolio founders; Masad 2026 published there.
• YC Root Access — YC’s recent-batch talk channel; multi-speaker batch sessions on a single technical theme. First wiki ingest from this channel is Garg 2026; Garg’s references suggest companion talks by Pete Koomen, Tom Blomfield, and Garry Tan in the same agent-harness-engineering session.
• Bookface — YC’s internal social platform + portfolio knowledge base (Tan built v1).
• GStack — open-source Claude Code harness toolkit; Tan’s recent project (Dangling first-mention).

Open questions

• YC AI-batch composition — what fraction of YC’s W26 / S26 batch is AI-native by default? First-party YC disclosure would substantiate the “this is the AI batch” implicit framing in both Tan and Hu.
• YC-portfolio adoption of GStack — Tan ships GStack; how many YC portfolio companies adopt it? First-party data.
• The 1-billion-startups-in-10-years analog — Hu’s 1,000× engineer thesis suggests dramatic founder-leverage shifts. Does YC’s accelerator throughput change in response?
• YC-partner-content cadence as a wiki source — if Tan + Hu are representative of YC’s content velocity, the wiki should expect to ingest more YC-partner-public-content sources at rate-of-batch-cycle (every ~3 months).
• The YC Root Access multi-talk session Garg sits in — Garg references “Pete and Tom and Gary” (likely Pete Koomen, Tom Blomfield, Garry Tan) as prior speakers. If the other slots in this session are ingestible, the wiki would gain a four-source YC Root Access session-level anchor on agent-harness engineering at startup scale — a candidate for its own thread page.