Buying behavior has changed. Before searching on Google, more and more buyers ask their question directly to an AI: "what natural skincare brand do you recommend?", "3 ethical fashion brands in France", "which project management tool for a small business?". ChatGPT, Perplexity, Gemini answer with specific names. Those who get cited get qualified traffic and credibility that classic SEO takes months to build.
The question is no longer "am I well-positioned on Google?" but "do AIs know me?"
Why your brand may not appear
Three technical reasons can partly explain why a site is missing from AI answers.
1. You don't have a llms.txt file
The llms.txt file is the AI equivalent of robots.txt. It explicitly presents to language models who you are, what you sell, and how to describe you. Without it, the AI rebuilds your identity from scattered fragments. With it, it has the answer directly.
2. Your content isn't readable by AI crawlers
A lot of sites load their content via dynamic JavaScript. ChatGPT and Perplexity don't execute JS during indexing: they read the raw HTML. If your catalog, your descriptions and your "About" page aren't in that static HTML, you're invisible.
3. You lack trust signals
AIs cite reliable sources. A site with no schema markup, no customer reviews, no structured "About" page looks like an anonymous page. The AI prefers to recommend what it clearly recognizes.
How to test it yourself in 3 minutes
Three tests to do in ChatGPT (free or paid version, doesn't matter). The goal isn't to appear in a generic list, it's to understand what the AI knows about you and in what context it might cite you.
Test 1 — sector awareness
What are the characteristics of a good [your sector] brand in [your country/region]?
This test reveals if the AI has a reference frame on your niche. If it cites competitors in the answer, you know who's already on its radar.
Test 2 — direct presence
Do you know the brand [your name]? What's their positioning?
Clear binary answer: it knows you or it doesn't. If it responds with wrong information, that's also an important signal — it means the data circulating about you is inconsistent.
Test 3 — guided recommendation
I'm looking for a [your country/region] brand specialized in [your precise specificity].
Do you know any?
Hyper-targeted query on your exact niche, not on the generic market. A brand well-positioned on its specificity has more chances to appear here than in a broad question about its sector.
What the results mean
If you appear in tests 2 and 3: existing AI presence. Focus on the consistency and richness of the information the AI returns.
If you only appear in test 2 (name search): the AI knows you but doesn't spontaneously recommend you. It's a content positioning problem, not an awareness problem.
If you don't appear at all: you're starting from zero in GEO (Generative Engine Optimization). It's common, and it's fixable.
The 3 levers to fix your AI visibility
Three technical signals explain the vast majority of absences in AI answers. No need to redo your site — first identify which ones you're missing.
A llms.txt file tailored to your brand
The llms.txt summarizes in a few dozen lines who you are, what you offer, and how to describe you to language models. It's the most direct signal to improve your presence in generative answers.
This file must be precisely tailored to your brand and consistent with the rest of your content. A generic or poorly structured llms.txt can even create inconsistencies that hurt your positioning.
Schema markup on your key pages
Schema markup (JSON-LD) structures your information in a format AIs understand directly. It covers your brand identity, your products, your commitments, your customer reviews. It's what allows an AI to cite you precisely on a specific query, not just on your name.
NanoLLM analyzes your site and generates a base of schemas tailored to your sector and actual content. Accessible data (description, logo, social networks, founding date, city) is integrated automatically. What can't be extracted from the HTML (product prices, hours, GPS coordinates) is tagged with precise instructions so you only have to complete the few missing fields before deploying.
A non-restrictive robots.txt
Some sites accidentally block legitimate AI crawlers. If your robots.txt blocks access to the right user-agents, you make yourself invisible — not because your content is bad, but because you accidentally shut the door.
Take action
Testing manually with ChatGPT gives you a snapshot. But this result varies depending on model version, language, and session context. A structured audit goes further: AI readability score across 30 criteria, priority actions ranked by impact, llms.txt generated for your brand, and JSON-LD schemas pre-filled from your actual data to complete and deploy.
Scan your site for free on nanollm.nanocorp.app. The report is generated in 60 seconds.