What is AI agent readiness?
AI agent readiness is how easily an AI system, from answer engines like ChatGPT, Perplexity, Gemini and Google AI Overviews to autonomous browsing and shopping agents, can crawl your site, understand its content, cite it as a source, and act on it. A site can rank well in classic search and still be nearly invisible to AI agents if its content is locked behind JavaScript, missing structured data, or blocked from AI crawlers. This scorecard measures the four things that decide it: access, machine-readability, extractability, and agentic action.
How the score works
The score is a weighted self-assessment across 20 checks in five categories. Each check earns full points for Yes, half for Partial and zero for No. Points sum to a 0 to 100 readiness score. Here is the full weighting, no black box:
| Category | Weight | What it measures |
|---|---|---|
| Crawler access & permissions | 25 | Can AI crawlers reach and read your pages at all |
| Structured data & machine readability | 25 | Is your content parseable through schema and semantic HTML |
| Content extractability | 20 | Can an AI lift a clean answer from your pages |
| Agentic actions & flows | 15 | Can an agent navigate and complete tasks on your site |
| Trust, identity & freshness | 15 | Will AI engines trust and attribute your brand |
Score bands: 0 to 39 not agent-ready, 40 to 64 emerging, 65 to 84 agent-ready, 85 to 100 agent-optimized.
The 20 checks, explained
1. Crawler access & permissions
- AI crawlers allowed in robots.txt. Your robots.txt permits the major AI user-agents (such as GPTBot, ClaudeBot, PerplexityBot and Google-Extended) rather than blocking them. If they are disallowed, you cannot be read or cited.
- llms.txt published. You publish an llms.txt file mapping your key content for language models, a low-cost signal that guides AI systems to your best pages.
- Content is server-rendered. Your important content appears in the raw HTML, not only after client-side JavaScript runs, because most AI crawlers do not execute JavaScript.
- No bot walls on legitimate agents. Your site responds quickly and does not CAPTCHA or hard-block well-behaved AI agents on public content.
2. Structured data & machine readability
- Schema.org JSON-LD on key pages. Organization, Product, Article and FAQ schema tell AI systems exactly what your pages are.
- Machine-readable pricing and specs. Prices and product details live in real text, not trapped inside images or PDFs.
- Semantic HTML. Correct headings, lists and data tables, so structure is explicit.
- Clean metadata. Every page has a unique, descriptive title, meta description and canonical URL.
3. Content extractability
- Answer-first content. The core answer sits in the first paragraph, easy for a model to lift verbatim.
- FAQ and Q&A blocks. Real buyer questions answered directly on the page.
- Clear definitions and entities. Your category, product and key terms are defined plainly.
- Comparison tables. Evaluative topics include tables an agent can parse and quote.
4. Agentic actions & flows
- Core info without a login wall. Docs, pricing and product details are reachable without authentication.
- Completable forms. Signup and contact flows have accessible labels an agent can fill in.
- Programmatic access. You expose an API, data feed or MCP endpoint for machine access.
- Clean, stable URLs. Human-readable, persistent URLs an agent can rely on.
5. Trust, identity & freshness
- Consistent facts on-site. Name, founding year and HQ are stated consistently across your own pages.
- Facts match third parties. The same facts align with LinkedIn, Crunchbase, G2 and similar sources.
- Visible freshness. Content shows publish and updated dates and is kept current.
- Third-party citations. Your brand is referenced on sources AI engines trust, such as review sites and press.
Why AI agent readiness matters now
Buyers increasingly research through AI systems before they ever reach your site, and machines now make up more of the audience than people do. If those systems cannot access, parse or trust your content, you are absent from the answer at the exact moment a buyer is deciding. The numbers behind the shift:
- Bots outnumber humans. Automated requests hit 57.5% of web traffic in mid-2026, the first time bots passed people online, driven largely by AI crawlers and agents (Cloudflare Radar).
- AI crawlers can't see JavaScript-only content. A Vercel/MERJ analysis of 500M+ GPTBot fetches found zero JavaScript execution, yet 57% of the top 1,000 sites reveal content only after JS runs (Lantern, ModPageSpeed). Google's Gemini is the main exception, via Googlebot's renderer.
- Nearly half of sites block an AI crawler. 44.9% of sites disallow at least one AI bot, often without realizing it costs them AI visibility (TechnologyChecker).
- Most on-page work isn't done. Only about 12% of sites use schema.org at all, and Product/Review schema is cited by AI far more often than generic markup, 61.7% vs 41.6% (SSRN study).
Being agent-ready is how you stay in the consideration set as discovery shifts from ten blue links to a single synthesized answer.
Frequently asked questions
What does AI agent readiness mean?
It is how easily an AI system, from answer engines like ChatGPT, Perplexity, Gemini and Google AI Overviews to autonomous browsing and shopping agents, can crawl your site, understand its content, cite it as a source, and complete actions on it. It spans access, machine-readability through schema and semantic HTML, answer extractability, and whether an agent can actually take action such as reaching pricing or completing a form.
How do I make my website ready for AI agents?
Start with access: allow the major AI crawlers in robots.txt and consider publishing an llms.txt file. Make content machine-readable with Schema.org JSON-LD, semantic HTML and real-text pricing. Write answer-first, add FAQ blocks and comparison tables, and keep entity facts consistent across your site and third parties. Finally, make sure core information is reachable without a login wall and that URLs are clean and stable. This checker scores all 20 of these factors.
Do AI crawlers read JavaScript-rendered content?
Mostly no. A Vercel/MERJ analysis of over 500 million GPTBot fetches found zero JavaScript execution, and most other AI crawlers behave the same way, so content that only appears after client-side rendering is invisible to them. Google's Gemini is the main exception, since it can use Googlebot's renderer. Server-side rendering or static generation is the safest way to guarantee AI agents see your content.
What is llms.txt and do I need it?
llms.txt is a proposed standard: a markdown file at your domain root that gives large language models a curated map of your most important content with clean links. It is a helpful signal rather than a strict requirement, and Google has said it does not use llms.txt for Search rankings, so treat it as a low-cost bonus. Here it is one factor within crawler access, not the whole score.
How is the AI agent readiness score calculated?
It is a weighted self-assessment. Twenty checks sit in five categories: crawler access and permissions (25 points), structured data and machine readability (25), content extractability (20), agentic actions and flows (15), and trust, identity and freshness (15). Each check scores full points for yes, half for partial and zero for no, summing to a 0 to 100 score. The full weighting is published above.