This translation has not been editorially reviewed yet. The German version is authoritative. Deutsche Fassung
GEO / AI
What does AI need to understand my company?
What an AI says about your business comes from text: from your website and from what others write about you. In order for it to clearly recognize you, the name, address, services and your profiles should be in fixed places and read the same everywhere: in the imprint, on an about-us page and in the structured data. Google recommends the information about the organization on the homepage or the About Us page; for its AI functions, no additional file is required. The llms.txt is a proposal whose benefits are open.
An AI recognizes your company by the same information
What an AI knows about your company, it is composed of pages that it has read during training or for an answer. If your website, company profile and directories contain different names, addresses or services, it may mix information or confuse you with another company. Google describes the information about the organization in the structured data exactly for this: it helps to understand the administrative information of a company and to distinguish it from others in the search results.
The imprint contains the information that can be checked
In Austria, § 5 of the E-Commerce Act requires name or company, the address of the establishment, contact details including e-mail address, if available company register number and company register court and the VAT identification number, for certain activities supervisory authority and chamber. This information is fixed and can be compared with registers. The same information belongs in the structured data: Google lists the legal name, address, telephone and e-mail, the UID number and other identifiers as characteristics for organizations. Write them there just like in the imprint.
The About Us page says who is behind the site
Google asks in its guide to helpful content whether it is clear to visitors who wrote a content, whether pages name an author where you expect it, and whether there is background to the author or the website, such as via an author page or the About Us page. Trust weighs the most after Google; The criteria for this are not a separate ranking factor. On the above-us side, who leads the company, since when it exists, where it works and what it offers, belongs to the people with their tasks. Google cites them as one of the locations for the information about the organization in the structured data.
Structured data: same information readable for machines
For the organization, Google does not know any mandatory information and recommends entering all those that apply to your company, such as name, legal name, address of the website, logo, address, telephone, e-mail and UID number. With sameAs you refer to pages of other websites that describe your company; Multiple references are allowed. the information must match the visible text; Google calls this explicitly for its AI functions. The structured data belongs in the source code that the server delivers. What is only created in the browser, the crawlers of the large AI providers do not see after an evaluation of Vercel.
What the audit of euid.com measures
In the GEO check area, the check reads whether the structured data on the organization or person contains references with sameAs, whether they give different names for the organization and whether a value in the schema contradicts the visible text. In the Data & Law check area, it checks the mandatory information in the imprint, including whether the UID number is valid. If the llms.txt is missing, this is an indication in the indexing without point deduction. If the file refers to pages that respond to 404 or invites crawlers that lock out your robots.txt, this is a finding in the file.
The llms.txt: a proposal with an open impact
The llms.txt has proposed Jeremy Howard 2024, the second version is from August 2026. The file is under /llms.txt, starting with the name as the heading, the only mandatory indication, and following with a short version and lists of important pages. It is intended for programs that search for information about a website on a request, less for training. AI providers publish llms.txt files for their own documentation; a crawler that reads the file does not name the side of the proposal. Google says that its AI functions do not require AI text files. The test file generates a llms.txt according to this proposal from setup and measurement, which you can download and drop. It costs little effort; It cannot replace imprint, over-us page and structured data.
- Write your company name the same everywhere: in the imprint, on the About Us page, in the Google company profile, in business directories and in the structured data.
- Create an About Us page that says who is behind the company, since when, where and with which services, and link them from each page.
- On the homepage, store organization-type structured data: name, legal name, address, telephone, email, logo, UID number and sameAs references to your profiles.
- Compare each item in the structured data with the visible text of the page.
- Download the llms.txt from your test file and drop it under /llms.txt if you want to use the proposal.
- Google allows multiple sameAs references. Suitable are pages that describe your company and on which you have influence yourself, such as your company profile, your page at a chamber or your entry at Wikidata.
- The UID number, if any, is in the imprint anyway. In the structured data, the field is called vatID.
- An author line with a link to a page about the person recommends Google where readers expect it.




