We Audited 49 Top YC Companies for AI Readiness. Nobody Scored an A.

The Experiment

AI agents are rapidly becoming the new interface layer of the internet. Shopping assistants, research tools, automated workflows — software agents are navigating the web on behalf of humans. But are the world's most valuable startups ready?

We used Inlay's AI readiness audit to scan 49 top Y Combinator companies — Stripe, Airbnb, Reddit, Coinbase, Notion, Figma, and dozens more. We measured everything: structured data, semantic HTML, llms.txt, MCP server availability, meta quality, AI crawler policies, and more.

The results were sobering. The average score was 59/100 — barely a C. Not a single company broke 80. The most innovative tech companies in the world, and none of them are doing AI readiness well.

Grade Distribution

The Full Leaderboard

All 49 companies, ranked by AI readiness score.

What the Leaders Get Right

Linear (76), Brex (75), and Notion (74) took the top three spots. A clear pattern emerged: developer-focused companies dominate the leaderboard. Linear, Vercel, Resend, Render — these companies build for developers, and they understand that machines need to read their sites too.

What they have in common:

• llms.txt files — Linear scored 82/100, Brex a perfect 100/100 on llms.txt
• Strong API discoverability — Linear hit 100/100, with well-documented, machine-readable API surfaces
• Clean meta tags — Top 5 averaged 85/100 on meta quality vs. 9/100 for the bottom 5
• AI-friendly crawler policies — All three scored 80/100 on AI crawler policy, welcoming rather than blocking agents

Notably, Notion was the only company in the top 5 with meaningful MCP readiness (73/100). Brex scored 30, and the other three scored zero. Even at the top, there's massive room for improvement.

The Failures

At the bottom of the leaderboard, some genuinely surprising names.

Reddit (30/100) — one of the internet's largest platforms — is nearly invisible to AI agents. Zero llms.txt. Zero API discoverability. Zero MCP readiness. Zero meta quality. A semantic HTML score of 9/100. Reddit's content is a goldmine, but AI agents can barely parse it.

Gusto (32) and Algolia (32) share nearly identical failure profiles: no llms.txt, no API discoverability score, minimal meta quality, broken semantic HTML. Algolia — a search company — is itself unsearchable by AI agents. The irony writes itself.

Coinbase (37) — a public company worth billions — earns an F. So does DoorDash (37). These aren't obscure startups. They're household names, and they're essentially invisible to the AI ecosystem.

What's Universally Broken

Some categories are strong across the board. Structured data averaged 89.7/100 — most companies have basic JSON-LD in place. Security & trust hit 80.5. These are table stakes, and companies know it.

But three categories are in crisis:

The message is clear: companies have nailed the basics of web presence — structured data, security headers, image optimization. But the AI-specific layer — the things that make a site readable by agents rather than browsers — is almost entirely missing.

Top 5 vs. Bottom 5: The Gap

The biggest differentiators between leaders and laggards aren't the traditional web metrics — it's the AI-native ones.

A 78-point gap on llms.txt. A 76-point gap on meta quality. The top companies aren't doing anything revolutionary — they're just doing the new basics that bottom-tier companies haven't even started.

AI Agents Are the New Browsers

The web was built for browsers. Then it was rebuilt for mobile. Now it needs to be rebuilt for agents.

When a customer asks an AI assistant “find me the best payroll software,” that agent crawls, parses, and evaluates websites. If your site is a JavaScript-heavy SPA with no structured data, no semantic markup, and no machine-readable content, you're invisible.

This isn't theoretical. AI-powered search — Perplexity, ChatGPT Browse, Google AI Overviews — is already reshaping how people discover products. Companies that aren't optimized for AI consumption are losing traffic today.

And the fixes aren't hard. The companies at the top of our leaderboard aren't using secret technology. They added an llms.txt file (30 minutes). They completed their meta tags (a few hours). They have structured data and clean HTML. These are quick wins with outsized impact.

The question isn't whether AI agents will become a major traffic source — it's whether your site will be ready when they do.