Introducing Visibian Research: how AI assistants recommend brands, in the open
Our research on brands in AI answers, plain explainers of how AI assistants work, and our company news. What we will publish here, and how.
Today we open visibianhq.com, the company site of Visibian. It is where we publish what we learn about how AI assistants describe and recommend brands, with the method and the sources set out in full. The product, the plans and sign-up stay on visibian.com.
Why publish it
When a buyer asks ChatGPT or Gemini which CRM, running shoe or dentist to choose, the answer names a few brands and leaves the rest out. Those answers reach a very large audience: on 28 October Google said its Gemini app has one billion users, and that questions asked in AI Mode are on average three times longer than ordinary searches (Google). Which brands make it into an answer, and why, is hard to see from the outside, and the answer keeps moving. SparkToro and Gumshoe.ai had volunteers ask ChatGPT, Claude and Google's AI the same twelve brand questions 60 to 100 times each, in November and December 2025. For ChatGPT and Google's AI, the chance of getting the same list of brands twice was less than one in a hundred (SparkToro, January 2026).
So one answer proves very little. A brand's place in AI answers shows in how often it is named over many runs, by which assistants, and in what words. Visibian measures that for its customers, and here we will publish it for whole categories.
What you will find here
- The monthly report. One fixed set of buyer questions in three categories (B2B software, online stores and local businesses), put in the same words to seven assistants, so one month can be read against the next. With it, a digest of the month's published studies, each figure with its author, date and link.
- Deep dives. One category at a time: which brands the assistants name, which comes first, how steady the answers are and where the assistants disagree. The first three are CRM for small sales teams, running shoes and family dentists.
- Studies. Data we collect ourselves, with the results for every site we look at. The first is what the top 1,000 websites let AI crawlers read.
- How AI works. Six explainers, from the first token to the final answer, built on real numbers from open models.
- News. Our company and product news, and the changes in AI search that matter to brands.
How we work
Every number on this site is traced to its source. Where a figure comes from our own measurements, we say which questions, which assistants and which dates. Where it comes from customer data, it is a total that no one can trace back to a customer. Where a method has limits, the transparency page says so.
Because a single answer has no fixed order, we report shares: how often a brand is named across many runs, and by which assistant.
You can follow the newsroom through its feed. For press enquiries, see the press center.