Service · Agent readiness
Agent Readiness: Make your site usable by AI agents
AI agents are software that open a site on a user’s behalf, read it, fill in forms, compare options and increasingly buy. Agent readiness makes your site readable and usable by them, so jobs like bookings, quote requests or purchases complete without stalling when an agent does them.
Agent readiness is the work of making a site readable and usable by AI agents. Key information is served in the initial HTML without JavaScript, forms use labelled, standard elements, prices and terms are written plainly, and agents acting for users are not blocked wholesale. New standards such as WebMCP are tracked while they are still in trial.
What does an AI agent do on your site?
Agents come in two forms. The first are assistants that open your page on a user’s behalf: OpenAI’s ChatGPT-User handles certain user actions in ChatGPT and Custom GPTsSource 1: Overview of OpenAI crawlers, and Anthropic’s Claude-User visits sites for user requestsSource 2: Does Anthropic crawl data from the web, and how can site owners block the crawler?. The second run a job end to end: pick a product, add it to a cart, check out.
Protocols for the second kind are emerging. Purchases inside ChatGPT run on the Agentic Commerce ProtocolSource 4: Stripe powers Instant Checkout in ChatGPT and releases Agentic Commerce Protocol codeveloped with OpenAI (29 Sep 2025) and Google’s checkout in AI Mode on the Universal Commerce ProtocolSource 5: About the Universal Commerce Protocol (UCP) and UCP-powered checkout on Google; both are US-only for now. Details on the E-commerce GEO page.
What can you do today?
| Area | What to do | Why |
|---|---|---|
| Content | Price, stock, terms and contact details in the initial HTML | The major AI crawlers do not execute JavaScriptSource 6: The rise of the AI crawler (Dec 2024); Google also recommends server-side rendering because not all bots can run JavaScriptSource 7: Understand JavaScript SEO basics |
| Forms | Labelled fields, real buttons, clear error messages | Agents fill forms by reading structure, not by looking at pixels |
| Flows | Simple, repeatable steps for bookings, quotes and contact | An abandoned flow is lost for an agent too |
| Access | Do not block agents acting for users wholesale; rate-limit abuse instead | ChatGPT-User and Claude-User arrive on user requestSource 1: Overview of OpenAI crawlersSource 2: Does Anthropic crawl data from the web, and how can site owners block the crawler? |
| Verification | Tell bots apart by user agent and published IP lists | OpenAI publishes its crawlers’ IP ranges as JSON filesSource 1: Overview of OpenAI crawlers |
Should you wait for WebMCP?
WebMCP is a proposed web API that lets sites offer structured “tools” to agents. The Chrome team opened an early preview on 10 February 2026, proposing a declarative API defined directly in HTML forms for standard actions and a JavaScript-based API for complex interactionsSource 3: WebMCP is available for early preview (10 Feb 2026). An origin trial started in Chrome 149 on 9 June 2026Source 8: Join the WebMCP origin trial (9 Jun 2026).
Tying a site’s core functions to a feature in trial would be a mistake. Our advice: build the basics above first, try WebMCP in a separate prototype and watch its standardisation.
What does Kaynaq do for agent readiness?
Scenarios
We define the jobs your customers would want an agent to do: book an appointment, request a quote, compare products.
Testing
We run those jobs with browser automation and with JavaScript disabled, and record where they break.
Fixes
We make content, forms and flows completable by agents.
Monitoring
We track agent visits and completed actions in server logs.
What we need from you, what we deliver
| What we need from you | What we deliver |
|---|---|
| The jobs you want agents to be able to do | Agent scenarios and test results |
| Access to a test environment or the live site | Form and flow fixes |
| Optional: access to server logs | Bot access and security recommendations |
| Contact with your developer | A WebMCP prototype assessment, if wanted |
Frequently asked questions
01Is it safe to allow agents?
An agent acting for a user should be treated like the user: payment and personal-data steps still require confirmation. You can verify bots by user agent and published IP lists and apply rate limitsSource 1: Overview of OpenAI crawlers.
02Is this about adding an AI chatbot to our site?
No. Agent readiness is about outside agents being able to use your site. Building a chatbot is a different job and outside our scope.
03Do we need WebMCP now?
No. As of June 2026 it is in an origin trial in ChromeSource 8: Join the WebMCP origin trial (9 Jun 2026). Getting the basics right makes more difference today.
Sources
- [1]Overview of OpenAI crawlersOpenAI · accessed: 26 September 2026
- [2]Does Anthropic crawl data from the web, and how can site owners block the crawler?Anthropic · accessed: 26 September 2026
- [3]WebMCP is available for early preview (10 Feb 2026)Chrome for Developers · accessed: 26 September 2026
- [4]Stripe powers Instant Checkout in ChatGPT and releases Agentic Commerce Protocol codeveloped with OpenAI (29 Sep 2025)Stripe · accessed: 26 September 2026
- [5]About the Universal Commerce Protocol (UCP) and UCP-powered checkout on GoogleGoogle Merchant Center Help · accessed: 26 September 2026
- [6]The rise of the AI crawler (Dec 2024)Vercel · accessed: 26 September 2026
- [7]Understand JavaScript SEO basicsGoogle Search Central · accessed: 26 September 2026
- [8]Join the WebMCP origin trial (9 Jun 2026)Chrome for Developers · accessed: 26 September 2026
How to cite this page
Kaynaq. (26 September 2026). Agent Readiness: Make your site usable by AI agents. https://kaynaq.pages.dev/en/services/agent-readiness
Updates and corrections
No corrections on this page yet. If you spot an error, write to us; we publish every fix here with its date.