Independent indexes of what AI models recommend.
AI Indexes publishes fixed-method measurements of what AI models say when a buyer asks: the products they name, how often, in whose favour, and with what reservations. Every index asks each question six ways, on behalf of three buyers, of the same models on the same day, judges every answer by one published method, and publishes the raw record behind every standing.
One instrument
Each index uses the same instrument: six framings of every buying question, three company sizes, twelve models, a judge that writes one label per product named with the sentence it read as evidence, and a calibration repeat that measures how much a standing moves with nothing changed. Only a move larger than that noise floor is reported as movement. The method, the judge rubric and the second-judge check are published in full on the method page.
Independence
No placement is for sale, no ratings are collected from anyone, and no index recommends a product. Each standing is derived from the raw record, which is published with it and free to download. The indexes do not claim the models can be steered, and a standing that moves between editions is reported as a measurement, not a result.
The publisher's interests are declared where they touch a standing. The publisher cofounded Gane, a product named in two GTM categories, and those pages carry the disclosure. No IT category contains a publisher interest.
To cite
Cite the index, not the publisher. Every edition of both indexes is published under CC BY 4.0.
GTM AI Recommendation Index, [edition], gtm-ai-index.com. Published under CC BY 4.0.
IT AI Recommendation Index, [edition], it-ai-index.com. Published under CC BY 4.0.
Contact
Corrections, claims of a vendor page and press, for either index: index@gtm-ai-index.com. AI Indexes publishes from Chandler, Arizona.