Agent readiness

01What it means

Agent readiness is a site's fitness for AI agents as visitors: can a plain, logged-out agent discover your content, read it in a form it can use, learn your policy for bots, and — at the frontier — act through the protocols you publish? It is a different discipline from SEO: search engines rank you; agents represent buyers. Ranking well and being legible to a buyer's agent overlap at the basics and diverge quickly.

It decomposes into a checklist. Discovery basics: robots.txt, a declared sitemap, real page metadata. Content agents can use: markdown negotiation, an llms.txt index. A stated bot policy. Then the agentic surface — MCP, A2A, signed bot auth — the emerging protocols that let agents act, not just read. The last group is honestly early: failing those checks today is normal; publishing them is the statement that agents are welcome.

We built the measuring stick and made our own site the pilot: fifteen weighted checks, scored 0–100, free at /tools/agent-ready. This site passes all fifteen, and three further agentic-commerce checks (product markup, declared feeds, ACP/UCP discovery files) report what they detect without scoring.

02Why it matters for buyers

• —Your next customer's first visit may be an agent: shopping agents already browse and compare vendors from whatever they can read — most businesses cannot even detect the visits.
• —It turns a debate into a number: a 0–100 score with fifteen named checks and copy-paste fixes beats "we should probably do something about AI search."
• —The work is unglamorous and compounding: structured content, clear terms, agent-facing signals — then re-check next quarter. That is most of the game.

03Common questions