Brainpan.AI measures how AI assistants describe brands — which companies ChatGPT, Gemini, Perplexity, Copilot and Claude name when someone asks for a recommendation, and which they leave out. The research on this page is published in full, including the underlying per-brand data, so any figure quoted here can be checked rather than taken on trust.
1. Media contact
Press enquiries are answered directly by the founder, not through an agency. Interview, data and fact-check requests are usually answered same day.
Kevin Walsh — Founder
Email: kwalsh@brainpan.ai
Telephone: +1-703-951-7195
For a correction request against a published figure, use the formal route in the AIVI™ methodology and correction policy rather than this address — it is logged and answered on the record.
2. Boilerplate
Written to be pasted without rewriting. Both versions are accurate as they stand; please do not combine them with claims from elsewhere on this site without checking.
Short (26 words)
Brainpan.AI measures how AI assistants describe and cite brands. It publishes the AIVI™ benchmark, a quarterly, reproducible study of brand visibility inside AI-generated answers.
Long (62 words)
Brainpan.AI is an AI visibility consultancy that measures how generative AI systems describe, cite and recommend brands. Its AIVI™ benchmark studies brand representation across ChatGPT, Gemini, Perplexity, Copilot and Claude using a declared prompt set and a published methodology, so results can be reproduced independently. The firm publishes its underlying data rather than summary scores, and operates a formal correction policy for challenged figures.
3. The research, and how to check it
The current published study is the AIVI™ Q2 2026 Insurance Benchmark. It analysed 1,742 organic citations across 134 brands. A separate study, 137 brands across five AI engines, found organic mention volume varying by up to 1.8x between engines — which is why a single “AI visibility score” is not a coherent metric.
Download and verify
Full benchmark report (PDF)
Top-10 results (JSON)
OpenAPI specification for the dataset
Methodology and correction policy v1.0
Per-brand JSON endpoints are published for every carrier in the cohort, linked from the benchmark page itself. A data desk can pull the figures directly rather than transcribing them.
4. What the data does not show
Stated here so it does not have to be discovered late in an edit.
It is not a quality ranking. Being mentioned more often by an AI assistant is not evidence that a brand is better, and we do not claim it is. Mention volume and recommendation are tracked separately because they diverge.
It is a snapshot, not a trend. Figures describe a defined measurement period against a defined prompt set. AI systems change between versions, and results move with them.
Retrieval mechanics are not documented by the vendors. We measure outputs. We do not claim to know how any specific product retrieves internally, and we are sceptical of anyone who does.
Comparisons hold only within a platform and mode. Not every AI surface exposes citations, so citation rates are comparable within a platform, not across them.
5. Reproducing figures and charts
Published research data may be quoted and reproduced with attribution to Brainpan.AI and a link to the source page. The full terms, including what may be redistributed, are set out in the research-data licensing section of our legal page.
If you need a figure in a specific format — a clean CSV of a table, a chart at print resolution, or a cut of the data we have not published — ask. We would rather supply it than have it retyped.
6. Spokesperson
Kevin Walsh is the founder of Brainpan.AI and the author of the AIVI™ methodology. He is available to comment on AI search measurement, generative engine optimisation, brand representation in AI-generated answers, and the limits of current AI visibility tooling.
A logo suitable for publication is available at brainpan-logo-transparent.webp.