More and more people no longer google — they ask. “Who builds AI solutions for SMEs in the Basel region?” someone types into ChatGPT, and gets a finished answer naming two or three providers. If your business is not among them, it does not exist for that person. That is the new reality, and it has a name: GEO — Generative Engine Optimization.
What GEO is, in one sentence
GEO is the practice of optimizing your website so that generative AI assistants — ChatGPT, Claude, Gemini, Perplexity — capture your content correctly and cite it in their answers.
The difference from classic SEO is in the result. A search engine returns a list of links from which the user chooses. An AI assistant formulates a finished answer and names individual sources within it. With SEO you compete for a spot on page one. With GEO you compete to appear in the answer itself — often just two or three mentions.
SEO and GEO are not opposites
A common worry: “Do I have to redo everything now?” No. The good news is that GEO and SEO largely overlap. A page that loads fast, is cleanly structured and has clear, fact-based content is readable for both Google and ChatGPT.
GEO adds a few specific signals:
| Signal | Purpose | SEO | GEO |
|---|---|---|---|
| Fast, clean page | Crawlers can read content reliably | ✓ | ✓ |
| Structured data (Schema.org) | Machines understand what a term means | ✓ | ✓ |
| Citable facts | AI can lift a clear statement | partly | ✓ |
| llms.txt | AI gets the core facts compactly | – | ✓ |
| AI crawlers allowed (robots.txt) | Bots may read the page at all | – | ✓ |
The four building blocks an SME needs
1. Citable facts instead of marketing prose
AI models prefer to lift clear, self-contained statements. “We are your innovative partner for the digital future” is worthless to an AI — it cannot cite anything from it. “Vollmer Labs is an AI studio in Reinach BL building AI agents for Swiss SMEs,” by contrast, is a statement an assistant can drop straight into an answer.
The rule: write at least a few sentences so they hold true and make sense without context. Those are the sentences that end up in AI answers.
2. Structured data (Schema.org)
Structured data is a machine-readable label in the source code that tells a machine: “This is a business, this is its address, this is a service, this is a frequently asked question.” For local SMEs the LocalBusiness or ProfessionalService schema is especially valuable, because it makes location, services and service area unambiguous. FAQ schema makes your answers to typical questions available directly as question-answer pairs — exactly the format AI assistants like to pick up.
3. An llms.txt file
llms.txt is a plain text file in your website’s root directory. It gives AI systems the key facts in compact form: what you do, which services you offer, verifiable details, contact. Think of it as a short dossier for machines. It is voluntary and no guarantee — but a clear, cheap signal that raises the chance of being reproduced correctly and completely.
4. Allowing AI crawlers in robots.txt
For an AI system to know your page at all, its crawlers must be allowed to read it. Bots such as GPTBot (OpenAI), the crawlers from Anthropic or PerplexityBot can be allowed specifically in robots.txt. It is a deliberate trade-off: you give up a degree of control over how your content is used and gain visibility in AI answers. For most SMEs that want to be found, the benefit clearly outweighs the cost.
Why we write about this — and do it ourselves
At Vollmer Labs we build websites so they are optimized not only for Google but also for AI search. That is not a paid extra, it is part of how we work. The very page you are reading has an llms.txt, structured data and deliberately allows AI crawlers.
That is the most honest proof we can offer: we apply GEO to our own site before selling it to you. If you need a website that shows up in the AI search of tomorrow, we build it exactly that way.
Where to start
If you take away just one thing: check whether your most important statements are citable — that is, true even without the surrounding text. It costs nothing, helps immediately and is the foundation everything else builds on. The rest — structured data, llms.txt, crawler permissions — we are happy to add.