Agentic Browsing in PageSpeed Insights: Checks and Fixes Google's PageSpeed Insights now includes an experimental 'Agentic Browsing' category in Lighthouse, which measures whether AI agents can understand and use a webpage. The checks focus on semantic HTML, accessible names, and stable layouts, but the category is not a ranking factor, and failed checks are not SEO penalties. Developers should prioritize fixing unlabeled controls and cumulative layout shift, while the optional llms.txt and WebMCP checks are low priority. Last updated: August 21, 2026 You opened PageSpeed Insights to check a slow page and noticed a new category: Agentic Browsing . Some checks passed. One says Not applicable . Others mention llms.txt and WebMCP. Should you worry? No. Agentic Browsing measures whether an AI agent can understand and use your page. It is not a Google ranking factor, and a failed check is not an SEO penalty. Here’s what the checks mean and which ones deserve your attention. Agentic Browsing is an experimental Lighthouse category in PageSpeed Insights and Chrome DevTools. Traditional Lighthouse checks ask: Is this page usable for people? Agentic Browsing asks: Can software understand this page and complete a task without guessing? That means identifying buttons, understanding form fields, reading page structure, and interacting with stable content. The category currently focuses on: llms.txt The score is based on applicable checks. A result marked Not applicable is not a failure. Important:Agentic Browsing is a readiness signal, not a ranking system. Agents, screen readers, and other automated tools rely on semantic HTML and accessible names. Avoid clickable