How Seenrix works

The check, step by step

We read the title and description of your homepage.
An AI model turns that into your brand name, your category and three questions a buyer would ask.
Each question goes to ChatGPT (OpenAI), Gemini (Google) and Claude (Anthropic) through their official APIs.
We record whether each answer names you, your position among the businesses named, and who else is named.

Sample size: 3 questions × 3 assistants = 9 answers, one answer per question per assistant.
Models: small, current models from each provider (for example gpt-5-mini, gemini-3.6-flash and claude-haiku-4-5). If one is unavailable we fall back to the next model from the same provider. Every answer in a report shows the model that wrote it.
Questions: written in English from your homepage. They never mention your brand. They include your city if you look local, and your country if you mainly serve one.
Brand matching: we compare a simplified form of your brand name and your domain name (lower case, letters and numbers only) with each business named in the answer, and with the answer text.
Position: your place in the order businesses are first named in that answer. 1 means you were named first.
Score: the share of answers that name you, out of the answers received.
Named instead: the number of answers that name each other business, with yours and the AI assistants themselves left out. The top eight are shown.

API answers can differ from the ChatGPT, Gemini or Claude apps, which may search the web or use your history.
Answers vary between runs. One check is a snapshot, not a ranking.
Brand detection is automatic and can miss unusual spellings, or match a different business with a similar name.
Nine answers is a small sample. Treat a change of one or two answers between checks as noise.

Through affiliate commissions on some paid tools we list. Details are on our disclosure page.