13 famous sites scored below · most fail

AI agents are your new visitors.
Most websites are invisible to them.

ChatGPT, Claude and Perplexity now browse, compare and buy on behalf of millions of people. If an agent can't read your site, you don't exist to it. We audit exactly what agents see — and ship the fixes.

Get your audit — $29 See how big sites score ↓

$29 flat · full fix package · delivered within 24h · money-back if you don't learn anything new

The Agent-Readiness Leaderboard

We ran our open-source scanner on well-known sites. Every score links to the full public report. Scanned 2026-08-28.

#SiteGradeScore
1 cloudflare.com A
92/100
2 shopify.com A
83/100
3 vercel.com A
82/100
4 stripe.com B
70/100
5 notion.so B
68/100
6 github.com B
65/100
7 huggingface.co C
51/100
8 airbnb.com D
48/100
9 anthropic.com D
48/100
10 nytimes.com D
47/100
11 booking.com F
28/100
12 reddit.com F
28/100
13 openai.com · blocks agents F
24/100

Methodology is public — scanner source on GitHub. "Blocks agents" means the homepage returns 403 to non-browser clients.

What we check

10 checks across discoverability, parsability and agent affordances — the same dimensions Lighthouse's new Agentic Browsing category cares about.

llms.txt

The emerging standard for giving agents a curated map of your site. Now part of Chrome Lighthouse’s Agentic Browsing audits.

AI crawler policy

Does your robots.txt welcome or block GPTBot, ClaudeBot, PerplexityBot and 10+ other AI crawlers — on purpose or by accident?

Structured data

JSON-LD tells agents what your business, products and prices actually are, instead of making them guess.

Content extractability

JS-only shells are invisible to most agents. We measure what an agent actually sees without a browser.

Agent affordances

MCP manifests, OpenAPI specs, security.txt — the interfaces that let agents act, not just read.

Meta, sitemap, feeds & speed

The boring fundamentals that decide whether agents can find, parse and trust your pages at all.

The fix package

Free scanners give you a score and leave. We give you the fixes, done.

$29 per site, one-time
  • Full audit report — every check, what agents currently see, prioritized by impact
  • llms.txt + llms-full.txt, written for your site — curated, spec-compliant, ready to upload
  • robots.txt AI-crawler policy — explicit rules for 14 AI bots, matched to your business intent
  • JSON-LD structured data — copy-paste blocks for your key pages
  • Agent-facing docs page — a /for-agents page that tells agents what you offer and how to act
  • Free re-scan — apply the fixes, we verify your new score
Get your audit — $29

Delivered within 24h. Email us your URL — you get a free score preview before paying anything.

Why now

Google added agentic-browsing audits to Lighthouse this year. OpenAI, Anthropic and Perplexity agents already do real purchasing and research for their users. The sites that are readable to agents get recommended; the rest silently drop out of the answer. Being early costs $29. Being late costs the traffic.