How to run a 30-minute agent readiness review with your team
In short
Invite four or five people, ask each to take the self-check alone beforehand and use the meeting to compare answers. Disagreements and not sure answers are the findings. Leave with three decisions: your position on AI bots, one readability fix to investigate and one named owner for agent policy.
Who and what beforehand
- The marketing, ecommerce or digital lead who owns the outcome, as chair.
- Whoever runs the website platform or its hosting.
- Whoever manages the CDN, firewall or bot protection.
- Someone from payments or fraud, if you sell or take bookings online.
- For publishers, whoever sets the policy on reuse of content.
Ask each person to take the self-check alone, without comparing notes, and to write down their level, their five area scores and the questions they answered no or not sure. Different answers to the same question are useful: if marketing says robots.txt welcomes AI crawlers and security says the firewall blocks them, you have found your first action.
The half hour
| Minutes | What happens | What you leave with |
|---|---|---|
| 0 to 5 | Name the one task a customer's agent should be able to do: buy, book, quote, subscribe or read. | A single task to judge everything against. |
| 5 to 15 | Go through the five areas and read out the questions where answers differ or are not sure. Do not debate fixes. | What is true today, or that nobody knows. |
| 15 to 25 | Turn every not sure into a question with a name and a date, using the sentences on ask your web team. | A list of open questions with owners. |
| 25 to 30 | Take three decisions and write them down. | Your position on AI bots, one readability fix, one named owner. |
The written position
One page is enough: which AI visitors you allow and refuse and why, how verified agents are treated at checkout or booking, who owns the policy and when it will be reviewed. A review date matters more than completeness because the rules keep changing.
- 6 Aug 2026
Cloudflare said on 6 August 2026 that a recipe site could charge a fraction of a cent per fetch and a local paper could license articles at read time without a licensing deal or login.
- 1 Sep 2022
RFC 9309 states that robots.txt rules are not a form of access authorization; enforcement needs bot management.
A scanner comes after
Run a scanner after the meeting, once you know which questions you want it to answer. Do not set a score as the goal: scores from different tools measure different things, and none of them measures whether customers' agents finish tasks on your site. The tools that check page names three free ones.
Related
- The self-checkEighteen questions scored in your browser.
- Ask your web teamThe sentences to send, grouped by area.
- The one-hour auditWhat a leader can check alone.
Sources
- Cloudflare: the agentic internet, WebMCP example (6 Aug 2026) (blog.cloudflare.com/the-agentic-internet), read 4 Nov 2026.
- RFC 9309 Robots Exclusion Protocol (rfc-editor.org/rfc/rfc9309), read 4 Nov 2026.
Last reviewed