What AI agents find on the Fortune 500

137 of 500 homepages gave our scanner nothing it could assess. Our scanner reads from a cloud server and says who it is, as many AI agents do.

• 33 refused every automated reader we tried, whether it presented itself as our scanner, a Chrome browser or OpenAI's GPTBot.
• Most of the rest let some readers in and turned others away, depending on which reader asked and where it was hosted.
• So this is not a measure of whether any AI agent can reach these sites. It shows that access depends on who is asking.

Few sites block AI crawlers on purpose: 4% of robots.txt files restrict OpenAI's GPTBot by name.

Structured data (JSON-LD) states facts such as who a company is and what it sells in a form an agent can read directly. Without it, an agent has to work those facts out from the page text. Of the 363 homepages we could read, 43% carry no JSON-LD in the page as fetched, and 51% don't name the company, a product or a service in it (no Organization, Product, Service or similar type).

• A census, not a sample: all 500 companies on the 2026 Fortune 500, using the website on each company's Fortune profile.
• Homepage only, fetched once per site without running JavaScript, by AgentReady’s free quick check.
• Where a bare domain could not be read, we tried the www address once more. Five investor-relations or regional sites were replaced by the company’s main site.
• Totals only. We do not publish company names or individual results.
• Sites change; these numbers describe one scan on the dates above.