Rendering without JavaScript
Your key pages must deliver their text in the initial HTML: server rendering or static generation. ChatGPT reads pages that are readable without JavaScript.
Technical GEO
Technical GEO: a website AI engines can read
Technical GEO makes your website readable and understandable for AI crawlers: pages rendered without JavaScript, schema.org structured data, an llms.txt file, a robots.txt open to GPTBot, ClaudeBot and PerplexityBot, and Bing indexing. Without this foundation, no content can carry your AI search optimization.
Updated
6
AI crawlers to check in your robots.txt
HTML
raw: what most AI crawlers read
Bing
the index ChatGPT and Copilot read
Claude Code
behind our in-house audit tools
The problem
Many websites are invisible to AI
A website can look great to a human and empty to an AI. The crawlers that feed ChatGPT, Claude or Perplexity do not behave like Googlebot: many read the raw HTML returned by the server, without running JavaScript. A site built as a single-page application (React, Vue, Angular without server rendering) then shows them an almost blank page.
Other sites block AI crawlers without knowing it, in their robots.txt or at firewall level. Others are missing from the Bing index, which ChatGPT and Copilot rely on. Technical GEO fixes these blind spots before anything else.
What we check
The technical points that matter in GEO
Schema.org structured data
Organization, Service, Product, FAQPage, Article, BreadcrumbList: they tell engines explicitly who you are, what you sell and at what price.
llms.txt file
A text file at the root of the site that presents your essential pages to language models, in a format that is simple to read.
robots.txt and firewall
We check that AI crawlers are blocked neither by your robots.txt nor by overzealous bot protection.
Sitemap and Bing indexing
Clean XML sitemap, Bing Webmaster Tools, IndexNow: your new pages quickly enter the index ChatGPT and Copilot read.
Speed and missing pages
Fast pages, and above all the missing ones: an offer, a city, a comparison that AI engines look for and cannot find on your site.
AI crawlers
The crawlers to allow in your robots.txt
Every AI company uses its own crawlers. Blocking them by default means disappearing from that engine’s answers.
| Crawler (user-agent) | Company | Role |
|---|---|---|
| GPTBot | OpenAI | Collects pages for OpenAI models |
| OAI-SearchBot | OpenAI | Real-time search for ChatGPT |
| ClaudeBot | Anthropic | Collects pages for Claude |
| PerplexityBot | Perplexity | Indexes sources for Perplexity |
| Google-Extended | Allows or blocks use of your pages by Gemini models | |
| Bingbot | Microsoft | Bing index, read by ChatGPT and Copilot |
Our tools
Audit tools built with Claude Code
Rather than stacking generic plugins, we built our own audit tools with Claude Code. They crawl your site the way an AI crawler would: they fetch the raw HTML, measure the share of text readable without JavaScript, check structured data, llms.txt, robots.txt and the sitemap, then spot the pages that are missing given your tracked prompts.
These tools evolve with our method, updated every week from the combined data of all our active clients. They let us audit a site in depth without spending weeks on it, and verify every fix once it is live.
Implementation
We fix it, or we guide your team
- 01
Technical audit
Our in-house tools run across the whole site, cross-referenced with the results of your GEO audit.
- 02
Prioritization
Each item is ranked by its expected effect on your business prompts and its implementation cost.
- 03
Fixes
Depending on your organization, Archipel makes the changes directly or writes precise instructions for your developer, in English or French.
- 04
Verification
We check every fix in production: HTML served, markup, crawler access, indexing in Google and Bing.
- 05
Tracking
The effect on your citations shows up in your AI visibility tracking, prompt by prompt.
llms.txt
llms.txt: useful, not magic
The llms.txt file is a proposed standard: a Markdown file placed at the root of the site that summarizes the business and lists the important pages for language models. We set it up as part of our GEO program, because it costs almost nothing and makes your site easier to read.
But an llms.txt replaces neither readable pages, nor GEO content, nor a presence in the sources AI engines cite. Be wary of anyone selling it as the answer to AI search optimization.
FAQ
Frequently asked questions
What is an llms.txt file?
An llms.txt file is a Markdown text file placed at the root of a website that presents the business and lists the essential pages for language models. It makes the site easier for AI to read but is not enough to get cited.
Should you allow GPTBot and ClaudeBot in robots.txt?
If you want to be cited by ChatGPT and Claude, yes: blocking GPTBot, OAI-SearchBot or ClaudeBot reduces your chances of appearing in their answers. Also check that your firewall does not block them.
Why is a JavaScript website a problem for GEO?
Because many AI crawlers read raw HTML without running JavaScript. A single-page application without server rendering looks empty to them; ChatGPT reads pages that are readable without JavaScript.
Does structured data help AI search optimization?
Yes: schema.org markup (Organization, Service, FAQPage, Article) explicitly describes your company and offers, which helps engines understand them and reuse them without errors.
Why is Bing important for GEO?
ChatGPT and Copilot rely on the Bing index. A page missing from Bing has little chance of being cited by them, even if it ranks well in Google.
Does Archipel change our website directly?
Depending on your organization, Archipel makes the technical changes itself or gives precise instructions to your developer, then verifies every fix in production.
Shall we get started?
Free GEO audit: your rank on ChatGPT, Claude and Google AI Overviews, your competitors and your first actions. Reply within half a business day.