Agentic engine optimization: Google AI director outlines new content playbook

Addy Osmani, a director of engineering at Google Cloud AI, published new guidance on Agentic Engine Optimization (AEO), a model for making content usable by AI agents.

He positioned this AEO (not to be confused with Answer Engine Optimization) as parallel to SEO, built for systems that fetch, parse, and act on content autonomously.

What he’s seeing. AI agents collapse multi-step browsing into a single request. They don’t scroll, click, or engage with UI — they extract what they need instantly. That makes most traditional engagement metrics irrelevant.

The token problem. Osmani highlighted token limits as a core constraint shaping content performance. Large pages can exceed an agent’s context window, causing:

• Truncated information.
• Skipped pages.
• Hallucinated outputs.

His takeaway: token count is now a primary optimization metric.

Content needs to change. Osmani recommended restructuring content for how agents read:

• Put answers early (ideally within the first ~500 tokens).
• Keep pages compact and focused.
• Avoid long preambles and buried insights. (Agents have “limited patience” for this, he noted.)

Markdown over HTML. He also recommended serving clean Markdown alongside traditional pages.

• Markdown reduces noise from navigation, scripts, and layout, making content easier and cheaper for agents to parse.
• This includes making .md versions directly accessible and discoverable.

Discovery and structure. Osmani pointed to emerging patterns for helping agents find and use content:

• llms.txt as a structured index of documentation.
• skill.md files to define capabilities.
• AGENTS.md as a machine-readable entry point for codebases.

These act as shortcuts for agents deciding what to read and use.

Why we care. This adds a new optimization layer alongside SEO. If agents can’t efficiently parse your content — due to token limits, structure, or format — they may skip, truncate, or misinterpret it. That directly affects whether your content is used, cited, or acted on in AI-powered experiences.

Between the lines. To be clear, the type of AEO Osmani discussed in his article is unrelated to Google Search or organic search ranking. Of note, Google’s John Mueller recommended against markdown pages and Google doesn’t use the llms.txt file.

• Osmani’s article highlights how AI systems interact with the web and what “optimized” content may look like in that environment.
• AEO shifts the goal from driving visits to enabling successful outcomes inside AI workflows.

The article. Agentic Engine Optimization (AEO)

Dig deeper:

• Meet llms.txt, a proposed standard for AI website content crawling
• llms.txt isn’t robots.txt: It’s a treasure map for AI
• No, llms.txt is not the ‘new meta keywords’
• Why LLM-only pages aren’t the answer to AI search
• Google’s John Mueller on SEO vs. GEO: Focus on audience behavior

View the full article