YC Companies · 10 min read

All YC AI Companies — Complete Database 2025

Short answer

YC has funded over 600 AI-related companies across its portfolio since 2005, with the vast majority concentrated in batches from 2019 onward and a dramatic acceleration from 2023. As of 2025, AI companies represent approximately 50-60% of each new YC batch — up from approximately 15% in 2021 and under 5% before 2019. This page provides a comprehensive sector-by-sector analysis of YC's AI portfolio, the most significant companies in each category, funding patterns, and what the complete YC AI database reveals about where the AI startup opportunity is most durable.

YC AI Portfolio at a Glance (2025)

Total AI-related YC companies: 600+
Proportion of recent batches: ~50-60% (W23, S23, W24, S24, W25)
Most valuable YC AI company: OpenAI (W14, ~$80B+ valuation as of 2024)
Second most valuable: Anthropic (founded by former OpenAI team, YC-connected)
Most valuable YC AI company from recent batches: Sierra AI (W24, $4B+ valuation)
Largest AI subcategory: Developer tools and infrastructure (~35% of AI companies)
Fastest-growing subcategory: AI agents (emerged as distinct category in S24)

The Answer Layer: YC AI Companies by Subcategory

Foundation Models and Core AI Research

The earliest and most valuable layer of the AI stack. YC funded the companies that built the foundational capabilities.

Company	Description	Batch	Status/Valuation
OpenAI	GPT model family and AI platform	W14	$80B+ (2024)
Scale AI	AI data labeling and annotation	S16	$13.8B (2024)
Weights & Biases	ML experiment tracking	S17	$1.25B
Cohere	Enterprise LLM provider	W23	$5B+
Together AI	Open-source AI inference	S23	$1.25B

AI Infrastructure and Developer Tools

The largest subcategory — companies building the picks-and-shovels of the AI ecosystem.

Company	Description	Batch	Notable
Braintrust	LLM evaluation and dataset management	W24	$36M raised
Helicone	LLM observability and monitoring	W24	Growing rapidly
Portkey	AI gateway and request management	W24	10,000+ developers
Langtrace	Open-source LLM observability	W25	Early stage
LanceDB	Vector database for AI	S23	Open-source leader
Weaviate	Vector search database	—	$50M raised
Trieve	Search and RAG infrastructure	S24	Early stage
LastMile AI	AI testing infrastructure	W25	Early stage
Composio	AI agent tool integrations	W25	$12M raised
Weights & Biases	Experiment tracking	S17	$1.25B valuation

AI Agents and Automation

The fastest-growing subcategory as of 2024-2025.

Company	Description	Batch	Notable
Cognition (Devin)	AI software engineering agent	S24	$175M, $2B valuation
MultiOn	AI browser automation agent	S24	$15.5M raised
Lindy	AI employee platform	S24	Significant traction
Induced AI	Autonomous web agent	S24	Active
Hyperwrite	AI writing and research agent	S24	Active
Lutra	AI workflow automation	S24	Active

Enterprise AI Applications

Vertical AI products for specific enterprise workflows.

Company	Description	Batch	Valuation/Status
Harvey	AI for legal professionals	W24	$1.5B+
Sierra AI	Enterprise conversational AI	W24	$4B+
Abridge	Clinical conversation AI	W24	$150M raised
Ambience Healthcare	Clinical documentation AI	W23	$70M raised
Anterior	Clinical decision support	W25	$20M raised
Tennr	Healthcare intake automation	S24	Active
Rogo	Financial analysis AI	W25	Active
Fintool	Investment research AI	W23	Active
Lexi	Immigration law AI	S24	Active
Decagon	Customer support AI	S24	Active
Nooks	AI sales dialer	S24	Active

AI for Developers (Coding and Engineering)

Company	Description	Batch	Notable
Augment Code	AI coding assistant	S23	$252M raised
Sweep	AI code review	S22	Active
Grit	AI code migration	W23	Active
Graphite	AI-assisted code review	W22	$20M raised

Consumer AI Products

Company	Description	Batch	Notable
Perplexity AI	AI answer engine	S23	$9B+ valuation
Captions	AI video creation	W23	Active
Dola	AI calendar assistant	W24	Active
Sana	AI learning management	S22	Active

AI for Indian and Emerging Markets

Company	Description	Batch	Notable
Sarvam AI	Indian language AI models	W25	$41M raised
Krutrim	Indian AI assistant	—	Unicorn
Bureau	AI fraud detection India	S23	Active
Docsumo	Document AI for financial data	W22	Active India

The Data Layer: YC AI Funding Patterns 2019-2025

AI Company Proportion by Batch Year

Year	Approximate AI % of Batch
2019	~8%
2020	~12%
2021	~18%
2022	~15% (pre-ChatGPT correction)
2023 (post-ChatGPT)	~45%
2024	~55%
2025	~60%

Post-Demo Day Fundraising by AI Subcategory (2023-2025 Batches)

Fastest fundraising (median 4-6 weeks to seed close):

AI agents with demonstrated task completion
Vertical AI with signed enterprise contracts
AI infrastructure with 1,000+ developer users

Slower fundraising (median 8-12 weeks):

Consumer AI without strong Day-30 retention
Horizontal AI "for everything" products
AI products directly competing with foundation model native features

Hardest fundraising:

AI wrappers without proprietary data or workflow integration
Consumer AI without a clear monetization path
AI products in categories where OpenAI/Anthropic has announced competing products

Most Valuable YC AI Companies by 2025

Company	Estimated Valuation	Category
OpenAI	$80B+	Foundation models
Scale AI	$13.8B	AI data infrastructure
Sierra AI	$4B+	Enterprise conversational AI
Perplexity AI	$9B+	AI search
Cohere	$5B+	Enterprise LLM
Harvey	$1.5B+	Legal AI
Weights & Biases	$1.25B	ML infrastructure
Together AI	$1.25B	Open-source inference

The Context Layer: What the YC AI Database Reveals About Durable Opportunities

Pattern 1: Infrastructure outlasts applications

Across YC's AI portfolio history, infrastructure and tooling companies have consistently outperformed application companies in durability. Scale AI (founded 2016), Weights & Biases (founded 2017), and LanceDB are all infrastructure plays that have remained relevant through multiple foundation model generations. Application companies are more exposed to commoditization risk when foundation models add native capabilities.

Pattern 2: Vertical AI outperforms horizontal AI

The most valuable recent YC AI companies — Harvey (legal), Sierra AI (enterprise support), Abridge (clinical), Anterior (clinical decision support) — are all deeply vertical. Their moat is domain data, regulatory relationships, and workflow integration — not AI capability per se. The horizontal "AI for everything" companies from equivalent batches have not performed comparably.

Pattern 3: The enterprise AI adoption lag

YC AI companies consistently underestimate enterprise AI adoption timelines. Companies that projected 6-month enterprise sales cycles in their applications have typically experienced 12-18 month cycles in practice. YC AI founders who adjusted their growth models for longer enterprise timelines — and who raised enough capital to sustain the longer cycle — survived and thrived. Those who projected faster adoption ran out of runway before closing enterprise deals.

Pattern 4: Proprietary data is the real moat

The most defensible YC AI companies have proprietary data that improves their product with usage and that competitors cannot easily replicate. Abridge's clinical conversation dataset, Harvey's legal reasoning dataset, and Perplexity's query-answer feedback loops are all proprietary data moats that compound over time. Founders building AI products should be able to name their specific data moat.

Keep reading

More on YC Companies

Founder Stories

Long-form, sourced stories from real YC founders.

Essay

Paul Graham: Do Things That Don't Scale.

Framework

The Mom Test — talk to users without lying to yourself.

Go deeper

Want the full data behind this answer?

Our YC database tracks 5,000+ companies, every batch, with application patterns, founder backgrounds, and pivot stories — the raw material we built this answer on.

Open the database →

FAQ

Frequently asked questions

How many YC companies are AI companies?

As of 2025, approximately 600+ YC portfolio companies describe AI as central to their product. This represents roughly 15-20% of YC's total portfolio of approximately 4,000+ companies. However, in recent batches (2023-2025), AI companies represent 50-60% of each cohort. The acceleration is dramatic: more AI companies were funded in W23-W25 than in all previous YC batches combined.

What is the most valuable YC AI company?

OpenAI, which went through YC's W14 batch, is the most valuable YC-affiliated AI company at an estimated $80B+ valuation following its 2024 funding rounds. Among companies from recent batches (2022-2025), Sierra AI (W24, enterprise conversational AI) and Perplexity AI (S23, AI answer engine) are among the most valuable at $4B+ and $9B+ respectively.

What categories of AI companies does YC fund most?

Infrastructure and developer tools is the largest AI subcategory in YC's portfolio — approximately 35% of all YC AI companies. Enterprise vertical AI (legal, healthcare, finance, sales) is the second largest at approximately 25%. Consumer AI is approximately 20%. AI agents is a newer but rapidly growing category at approximately 10% of recent AI batches.

How has the type of AI companies YC funds changed over time?

Before 2022, most YC AI companies were ML-augmented SaaS products — traditional software with AI features added. From 2023, AI-native products — where the AI is the core value delivery mechanism, not a feature — became the dominant pattern. From 2024, AI agents — autonomous systems that take multi-step actions — emerged as a distinct and growing category. The evolution reflects both foundation model capability improvements and the startup ecosystem's growing sophistication in building on top of them.

What makes a YC AI company defensible against foundation model commoditization?

The four most durable moats across YC's AI portfolio: proprietary data that improves the product with usage (Abridge, Harvey, Scale AI), deep workflow integration that creates switching costs (Sierra AI, Anterior), domain-specific fine-tuning on regulated or specialized datasets (Harvey's legal reasoning, Abridge's clinical conversation), and distribution advantages that give access to customers foundation model companies cannot easily serve directly (enterprise procurement relationships, healthcare system integrations).

Which YC AI companies are most relevant for Indian founders to study?

Sarvam AI (W25, Indian language models) is the most directly relevant for Indian founders building for Indian-language users. Bureau (S23, fraud detection) and Docsumo (W22, document AI) are relevant for Indian fintech and financial services founders. For founders building general AI products for the Indian market, the most useful study is how Sarvam AI framed the Indian language model opportunity in a way that was compelling to YC partners — making the case that India's linguistic diversity is a large, specific, addressable opportunity rather than a localization challenge.

What AI subcategories does YC's portfolio suggest are most saturated?

Based on the volume of similar companies in recent batches, the most crowded AI subcategories in YC's portfolio are: AI meeting notes and summaries (Fathom, Otter, and many others), AI writing assistants (competed away by ChatGPT native features), general-purpose AI chatbots, and AI-powered code review without specific workflow integration. These categories show the pattern of too many similar products competing for the same user and the same investor attention.

How do YC AI companies typically price their products?

Three primary pricing models dominate YC's AI portfolio. Per-seat SaaS with usage limits (most common for productivity tools — Harvey, Fathom, Abridge). Usage-based pricing on API calls or tokens processed (most common for infrastructure — Portkey, Helicone, LanceDB). Outcome-based pricing — charging for tasks completed or results achieved (emerging for agent products). The trend in 2024-2025 is toward outcome-based pricing for agent products as it better aligns with the value delivered by autonomous systems.

What retention benchmarks do successful YC AI companies maintain?

Based on disclosed data from successful YC AI companies, the benchmarks that correlate with strong Series A fundraises are: Day-7 retention above 40% for consumer AI products, monthly logo retention above 90% for enterprise AI products, net revenue retention above 110% (indicating expansion) for B2B AI SaaS, and task completion rate above 85% for AI agent products. The retention bar for AI products has become progressively higher as investors have applied lessons from S23's "retention test batch."

What is the YC AI database most useful for when preparing a YC application?

Three specific uses. First, benchmarking your AI product's retention and revenue metrics against funded companies at equivalent stages — the YC-Insights database provides these benchmarks by subcategory. Second, identifying the competitive landscape — which AI companies YC has already funded in your specific vertical tells you both the investor interest level and the products you will be compared to. Third, studying how successful AI companies described their proprietary data moat, workflow integration, or domain expertise in ways that convinced YC partners their product was defensible against foundation model commoditization.

← all YC Q&A

An independent resource · Not affiliated with Y Combinator · Last updated 2026-08-04