Agent Experiences

Optimize for AI agents

If you run a website, an online shop, or a software product, AI tools now visit on your customers’ behalf. Assistants like ChatGPT, Gemini, and Claude answer questions about you. AI agents go further and book, buy, or fill in forms.

Agent Experiences shows you how to design for both: assistants that answer and agents that act. Most of the site is about AXD (agentic experience design): designing how people and AI agents work together on a job, including what the agent may do, what the person sees, and how they stay in charge. The rest covers AXO (agent experience optimization): making your website easy for AI tools to find, understand, quote, and use.

Agent Analyzer

Find out what an agent can use on your page

The Agent Analyzer runs 43 checks: 31 on the page you give it and 12 on your site as a whole. They test whether an agent can load the page, read it without running JavaScript (code many sites use to build the page inside the browser), tell your content apart from your menus, see who wrote it, and take action on it. Each check tells you what the problem costs you today and what changes once it passes.

Four problems hold the score down. Your robots.txt (the file that tells bots which pages they may visit), your server, or a noindex tag (a line asking search engines not to list the page) turns AI tools away. Your content only appears after JavaScript runs. You have no usable robots.txt. Or your robots.txt doesn’t say which AI crawlers (the bots AI companies send to read websites) may use your pages for search, training, or fetching for someone. In that case each AI company’s defaults decide for you.

97/ 100Excellent

agentexperiences.com/blog · October 3, 2026

Structure20/20
Trust19/20
Technical14/15
Agent accessibility15/15
Schema markup matches the page: partly. The labels list 17 article headlines, but the main heading is “AXO Blog”.
Cites outside sources. No links to outside sources in the article content.

A real scan of this site, not a mock-up. The Analyzer now checks 35 of the 56 items on our audit checklist.

Agent Visualizer

Watch where the agent stops

It draws the same scan as the path an agent takes through your page. The agent arrives, reads, separates your content from the menus and footer around it, works out what the page is about, finds who wrote it, looks around, and tries to act. At each step you see what it found, with your headings as you wrote them and your form fields by name.

If the agent gets stuck, every step after that is marked as not reached, never as passed. An agent that couldn’t read your page never saw your structured data (labels in a page’s code that state facts such as the author, date, or price), so it can’t have judged it either way.

  1. 1Arrive

    Nothing in the way

    Load the page, and be allowed to use what is on it.

    Title
    Blog | Agent Experiences
    Language
    en

    The agent loads the page and is allowed to use it. Every later step depends on this one working.

  2. 2Read

    Nothing in the way

    Pull the words out of the page’s code, without opening it in a browser.

    Readable words
    923 in the page
    In the main content area
    862 words

    The words can be read straight from the page’s code, without a browser, so every agent gets the same text a person sees.

  3. 3Separate

    Nothing in the way

    Tell your content apart from the menus, header, and footer.

    Page sections (header, main content, footer)
    <main>, 17 <article>, <nav>
    Headings on the page
    h1 AXO Blog · h2 Why an Agent That Works 80% of the Time Finishes a Third of the Job · h2 The Same Agent, the Same Page, 34 Points Apart · h2 Muse and Dots Both Drive a Browser · h2 GEO, AEO, LLMO: What Do You Actually Call This? · h2 Atlas Is Dead, Agentic Browsing Isn't

    Your content stands apart from the menus and the footer, so a quote comes from the article and not from a menu.

An agent walks the whole path with nothing in its way.

The first three steps of a real walk through agentexperiences.com/blog, taken on October 3, 2026.

Field Notes

Notes on how AI agents read, quote, and act on the web, each with the date it was written. They cover new research, changes at the big AI platforms, and close looks at real websites, and every note says what to do about it.

All Field Notes

Get the monthly newsletter

Once a month we send what changed in AI search, in crawler policy (the rules websites set for which AI bots may visit), and in the technical standards that let agents connect to websites and apps. It comes with every short news post (Field Note) we published that month. No spam, and you can unsubscribe anytime.

We store your email address only to send you this newsletter. See our privacy policy.

How does your site score?

The Agent Analyzer fetches one of your pages the way an AI agent does and scores what it finds. You can check one page free every week.