AI SEO audit: what to check so AI search can find you, and how to run it with an agent
Fix access first, content second, and structure third, because a page no bot can reach gets cited by nobody.
A plain-text version of this site for AI assistants is at https://contengi.com/llms.txt, with the full product documentation in one file at https://contengi.com/llms-full.txt. Every page is also available as markdown by adding .md to its URL.
TL;DR
- An AI SEO audit checks two things: whether AI search engines such as ChatGPT, Perplexity, and Google's AI Overviews can reach, read, and cite your site, and whether an AI agent can run the classic SEO audit for you.
- Start with access. A page has to be indexed and eligible for a snippet to appear in Google's AI features, and OAI-SearchBot and PerplexityBot need to get past your robots.txt and your firewall.
- Check the Search generative AI setting in Search Console, which rolled out to every site in August 2026. An excluded site gets no traffic or impressions from AI Overviews or AI Mode.
- An agent with Search Console access finds what a crawler can't: pages with impressions and no clicks, keywords landing on the wrong page, and queries typed by machines.
- Skip the hacks. Google Search ignores llms.txt, and you don't need to rewrite content in a special way for AI.
Every audit tool now has "AI" stamped on the box. Some check whether ChatGPT can see your site, some feed a crawl report to a model and call it an audit, and some do both. The label tells you very little about which one you're buying.
This post splits the two jobs apart, lists what an AI SEO audit should check, shows how to run one with an agent, and shares what ours turned up on contengi.com.
What is an AI SEO audit?
An AI SEO audit is a review of whether your site can be found, read, and cited by AI search, run partly or fully by an AI agent. The first half is about access, the second is about who does the work.
Think of an MOT. One part checks the car is road-legal, the other is the mechanic and their kit. You need both, and a shiny diagnostic machine can't pass a car with no brakes.
What should an AI SEO audit check?
Start with access, then content, then structure. Fix them in that order, because a brilliant page that no bot can reach gets cited by nobody.
| Check | Why it counts | How to check it |
|---|---|---|
| Indexed and snippet-eligible | Required before a page can show as a link in AI Overviews or AI Mode | URL Inspection in Search Console |
| Search generative AI setting | An excluded site gets no traffic or impressions from Google's AI features | Search Console, Settings, Search generative AI |
| robots.txt for AI search bots | Blocking OAI-SearchBot keeps you out of ChatGPT search answers | Read your robots.txt line by line |
| CDN and firewall rules | Bot protection can block a crawler your robots.txt allows | Your CDN's bot settings and firewall logs |
| Rendered text | Content that only appears after scripts run can go missing | Compare what a bot receives with what you see |
| Structured data | Markup has to describe what's visibly on the page | Compare the schema with the page copy |
| Internal links | Pages nothing links to are hard to find | A crawl or the Search Console links report |
| Content quality | A unique point of view does more for AI visibility than any tweak | A page-by-page content review |
Access checks
To show up as a supporting link, a page must be indexed and eligible to be shown in Google Search with a snippet, and Google asks site owners to make sure crawling is allowed in robots.txt and by any CDN or hosting setup. There are no extra technical requirements beyond that.
Then open Search Console. The Search generative AI control rolled out to all websites worldwide on 31 August 2026. Including your site is the default, but if anyone on your team clicked exclude, you won't receive any traffic or impressions from those features.
Outside Google, each assistant runs its own crawler. Sites that opt out of OAI-SearchBot won't be shown in ChatGPT search answers, and GPTBot, the training crawler, is controlled separately. PerplexityBot surfaces and links websites in Perplexity's search results, and Perplexity's own docs walk through allowing it past a firewall. The robots.txt for AI guide has the lines to copy.
Content and structure checks
Render your key pages the way a bot sees them with the AI bot view, then read them as a human. On the content side, unique, useful content will likely influence your presence in generative AI search more than any other suggestion. A full content audit is where that review lives.
How to run an AI SEO audit with an agent
An AI SEO agent earns its place in the data-heavy half. Give it Search Console and a list of rules, and it does in minutes what takes a person a day of spreadsheet work. If you haven't set one up yet, here's how to build an SEO agent.
- Pull 90 days of Search Console by page and query. Use the API so nothing gets cut off at the export limit.
- Flag pages with impressions and no clicks. High impressions plus a deep average position means Google keeps testing the page and passing over it.
- Find keywords landing on the wrong page. Ask the agent which queries show a different URL from the one written for them.
- Filter out machine-shaped queries. Paragraph-length searches are usually visibility tools testing prompts, and they distort every average.
- Run the access checks. Fetch robots.txt, compare it against the bots above, and render your top 20 pages.
- Decide page by page, yourself. Refresh, merge, redirect, or leave alone. The agent proposes, you approve.
What our own audit found
Running this on contengi.com turned up less pages with technical problems than we expected and more with content problems.
One guide had collected more than 8,000 impressions in 90 days at an average position around 56, and earned a single click. Searches for "agentic SEO" mostly landed on our platforms roundup, while the page written for that exact term barely appeared at all, which is a textbook case of two pages competing for one question. And a few of our "queries" were whole marketing personas typed into Google as one search, which no founder does.
None of that shows up in a crawler. All of it showed up in an afternoon with an agent and Search Console, and each finding turned into a concrete job: one refresh, one merge-or-differentiate decision, and one rule to filter machine queries from every future report.
What can you skip?
Special AI files and formatting tricks. Google Search doesn't use llms.txt files to decide what appears in its AI features, and it doesn't need content chopped into tiny chunks either. An llms.txt file can still be useful for other assistants, it just won't move anything in Google.
Once access and content are sorted, track the result over time with LLM tracking so you know whether the fixes changed what the assistants say.
Frequently asked questions
What is an AI SEO audit?
An AI SEO audit checks whether AI search engines can reach, read, and cite your site, and uses an AI agent to do the data-heavy parts of the review. It covers access, rendered content, structured data, internal links, and content quality.
How do I check if ChatGPT can see my website?
Check that your robots.txt allows OAI-SearchBot and that your CDN or firewall isn't blocking it. Sites that opt out of OAI-SearchBot don't appear in ChatGPT search answers.
Do I need an llms.txt file for AI SEO?
Not for Google, which ignores it in Search and its AI features. It can help other assistants, so add one if it's cheap for you, but don't expect it to change your Google visibility.
Can an AI agent do a full SEO audit?
It can do the data work: pulling Search Console, spotting weak pages and cannibalisation, and checking access. Deciding what to refresh, merge, or redirect should stay with a person.
How often should I run an AI SEO audit?
Run the access checks monthly, because robots.txt and firewall changes break things without warning. A full content review every quarter is enough for a small site.