How to brief an agency on agent readiness
In short
Brief an agency on agent readiness with one page that asks for a fixed-scope, checkable proposal instead of an open retainer. Name the task, require a before scan from a public scanner, ask for each fix to be named by identifier, price it per fix or sprint, and require the same scanners and one agent journey after delivery.
What goes on the one page
- The main task an agent should be able to complete, and how you will know it did. Buy, book, request a quote or read an article.
- The starting evidence: a scan from at least one free public scanner and screenshots of what a plain fetch of your key pages returns.
- The standard to work to. Ask for each fix to be named by its AgentReady identifier, such as AR-FIND-01, or by the scanner's check name.
- The price: a fixed fee per fix or per sprint, not a monthly retainer for AI visibility.
- The proof: the same scanners re-run after delivery, plus one agent journey on the main task.
- Your platform: named experience with Shopify, WordPress, Next.js or whatever you run.
Why this shape works
The evidence in the list below says that only two changes stopped an agent from getting an answer in Ora's experiment: an answer hidden in JavaScript and a 403 returned to agent user agents. Both are web team fixes that a scanner can show before and after. A proposal that does not start there should explain which check each month of work will fix.
The same evidence is why a leader can ask for identifiers. The list of what a site should implement has stable names, so an agency can say which items it will deliver and you can check them without a developer.
Red flags in a proposal
- A promise of a position in AI answers. No agency controls what an AI product recommends.
- No before and after evidence from a scanner you can run yourself.
- A retainer that starts with content volume rather than the two blockers.
- A file such as llms.txt as the headline deliverable. In Ora's experiment it did not change whether an agent got the answer, although it can be a reasonable small addition.
What the evidence says
Give a leader a one-page brief that gets a fixed-scope, checkable proposal instead of an open retainer.
- 4 Nov 2026
Ora's scan is free, with no key, card or signup, and rate-limited by IP (read 4 Nov 2026).
- 19 Oct 2026
Ora's 19-configuration experiment (190 probes) found only two changes stopped retrieval: an answer hidden in JavaScript (0 of 5 for plain fetch) and a 403 to agent user agents (0 of 5 for both clients).
- 19 Oct 2026
AgentReady requirements carry stable identifiers such as AR-FIND-01 and strengths MUST, SHOULD and MAY; the spec itself scores no one.
- 19 Oct 2026
The AgentReady essentials for every site are: allow agents in robots.txt and bot protection, put the answer in the initial HTML, return real HTTP status codes, and list pages in a sitemap.
- Read 4 Nov 2026
Cloudflare's free scanner at isitagentready.com checks discoverability, content accessibility, bot access control, protocol discovery and commerce.
- Read 4 Nov 2026
is-agentic.com, built by Vercel with Ora, weights essential checks for content sites and excludes checks that do not apply.
Related
- What to ask a vendorOra first, then three others, scored on the same five questions.
- Ask your web teamThe sentences to send, grouped by area.
- Tools that checkThree free scanners, and the one we recommend.
Sources
- Ora pricing (JSON) (ora.ai/pricing), read 4 Nov 2026. Vendor claim.
- AgentReady.org guide (agentready.org), read 4 Nov 2026. Vendor claim.
- Introducing AgentReady.org (19 Oct 2026) (ora.ai/blog/agentready-launch), read 4 Nov 2026.
- Is Your Site Agent-Ready? (Cloudflare, isitagentready.com) (isitagentready.com), read 4 Nov 2026.
- Is Agentic: AI Agent Readiness Score for your Site and App (is-agentic.com), read 4 Nov 2026.
Last reviewed