Google Published Its AI Search Guide. The Headline Missed the Signal.
Reading the May 15 doc for what's true, what's positioning, and what's coming next.
Google’s Search Central team quietly published a new document on May 15. It’s called Optimizing your website for generative AI features on Google Search.
Most of the coverage zeroed in on one line. From Google’s perspective, AEO and GEO are still SEO.
That line is true. It’s also not the interesting part.
The interesting part is at the end of the doc, in the section nobody is quoting.
I’ll get there. First, what Google actually said…
What’s in the doc
The guide is focused and short. It covers five things:
How generative AI search actually works inside Google
Foundational SEO practices that still apply
Local business and ecommerce details
A “mythbusting” list of tactics you can ignore
Initial guidance on AI agents accessing your site
The headline framing is that AI search runs on the same systems as regular Search. Retrieval-augmented generation (RAG) pulls from the existing index. Query fan-out generates related queries to find supporting content. The ranking signals are the same.
If you’ve been working in this space, none of that is news. The novelty is that it’s now in official Google documentation. The mechanics that AEO consultants have been talking about for the last 18-24 months are confirmed by the source.
The mythbusting section is the operator gift
The most useful part of the doc is the list of things Google tells you to ignore.
llms.txt files and AI-specific markup
Chunking content into small pieces for AI
Rewriting pages specifically for AI systems
Chasing inauthentic mentions across the web
Overinvesting in structured data for AI purposes
If you’ve been pitched any of these as required work for AEO, this doc is your refund document.
I’ll say what most won’t. There’s an industry of agencies and consultants who built AEO service lines around exactly these tactics. The economics of selling a separate AI optimization retainer depend on AI optimization being meaningfully different from SEO. Google just confirmed it isn’t, not from their perspective, and not for their surfaces.
The principles haven’t changed. Create unique, non-commodity content. Make it crawlable. Use clear structure. Don’t manipulate. These are SEO fundamentals. They are AEO fundamentals too. They were never different.
What did change is the bar
Saying AEO and SEO are the same discipline is not the same as saying nothing changed.
Commodity content used to rank. It doesn’t get cited.
A “10 tips for first-time homebuyers” post performed fine in 2022. It doesn’t show up in an AI Overview in 2026. The model has a corpus of those posts. It synthesizes one. Nobody gets the citation.
What gets cited is the post titled something like “Why we waived the inspection and saved money: a look inside the sewer line.” That’s not my example. It’s directly from Google’s doc. First-hand experience. Specific stakes. A perspective the model can’t generate from common knowledge.
The principles are the same. The bar is higher.
If your content operation hasn’t adjusted for that, no AEO tactics will save you. Not chunking, not schema, not llms.txt. The model reads the substance.
The section nobody is reading
Now the part that matters.
Near the end of the doc, in a section called “Explore agentic experiences,” Google describes something most operators are not paying attention to.
AI agents are starting to access websites the way browsers do. They analyze screenshots. They inspect the DOM. They interpret the accessibility tree. They take actions, not just read content.
Google references the Universal Commerce Protocol (UCP), an emerging standard for letting Search agents do more than retrieve. The doc points readers to the web.dev guide to agent-friendly website best practices.
This is a phase transition, not an extension.
For the last fifteen years, optimization meant “will my content rank and convert when a human reads it?” That work isn’t going anywhere. Most pages still need to do exactly that. But a growing share of traffic is going to be agents acting on behalf of humans. The optimization question for those agents is different.
It’s “can this agent complete the user’s job here?”
That’s a question about interface, not content. About task completion, not engagement. About a machine reader, not a human one.
A few things it changes in practical terms.
Form fields need to be machine-parseable. Not just labeled, but structured so an agent can map intent to action.
Product pages need clear, extractable specs. Not just compelling copy.
Pricing pages need explicit, accessibility-tree-readable structure. Not just visual hierarchy.
Booking flows need to work without JavaScript-heavy patterns that browser agents trip on.
None of this is exotic. It’s standard accessibility work, plus an awareness layer about what agents need to act.
But here’s the operator point. Nobody is doing this yet. There are no UCP service lines from agencies. There are no “agent optimization” frameworks circulating on LinkedIn. There are no benchmarks for agent task completion on the average enterprise site.
The space is empty and that’s the gap.
What to do now
Three things.
First, read the Google doc yourself. It’s short and clearly written. Don’t take anyone’s summary as the full thing, including this one.
Second, audit your content production against the non-commodity bar. If your last six posts could have been written by a model with no context about your business, they aren’t going to get cited. Fix that before you touch anything else. The model already knows what everyone else knows. Publish what only you know.
Third, start thinking about your site as a place where agents act, not just where humans read. Walk through your highest-intent pages and ask: if an agent had to complete the buyer’s job here, could it? If the answer is no, that’s your roadmap.
The takeaway
The doc Google published is the most useful thing they have put out on this topic.
The “AEO is SEO” line is the bait. The mythbusting list is the gift. The agent section is the signal.
Most of the industry will spend this week arguing about the bait.
Spend it on the signal.
Get your site agent ready 😬