ARO: agent readiness optimization

AI agents don't just read your site; they try to use it: find the price, fill in the form, book the slot. ARO is making sure they can.

Two kinds of agent

Some agents fetch HTML and never run JavaScript. Others drive a real browser. A site is agent-ready when it works for both, so we check every page twice, once with JavaScript off and once with it on. The gap between the two is the first thing we look at.

What we check

• The JavaScript gap: how much of the content and navigation exists only after scripts run.
• Real links and buttons: navigation is <a href> and actions are <button>, not clickable divs.
• Labeled controls: every form field has a name an agent can read.
• Walls: cookie banners and pop-ups that block content until someone clicks.
• Bot challenges: CAPTCHAs and challenge pages on the pages an agent needs.
• Time to content: how long until the main content exists on the page.
• The accessibility tree: landmarks, roles and names, the structure many agents navigate by.
• A real task: whether each mode can follow the main navigation and finish one representative journey.

Run it yourself

The step-by-step version is in How to audit your site's agent readiness. The free AI Agent Readiness check (my own product) looks at the structural signals on a page in one pass. It's a heuristic and doesn't run a live agent.