AI Indexes
AI IndexesPUBLISHERSubscribe
Licensing

License the catalog.

The category map and product table behind the four AI Indexes, as data: 363 categories, 18,679 products from 7,998 vendors with their websites, 8,052 product descriptions, and what fourteen AI models answer in every category, refreshed every month.

363CategoriesAcross 27 buying functions in four indexes: GTM, IT, HR and Finance.
18,679ProductsEvery product the models named, each spelling folded to one name; 12,358 tied to their own website.
7,998VendorsThe companies behind the products, each with its own website; 1,515 make more than one product here.
8,052DescriptionsOne paragraph each, from the product's own site and the answers that name it.
17,200Spellings resolvedNames as the models wrote them, each read to the product it means.
91,476Answers a monthFourteen models, six framings, three buyer sizes, in every category. October 2026 Edition: 472,606 labels.
84.6 million wordsof AI answers read and judged since September 2026: 170,824 answers, 495 words each on average, about 940 novels' worth. Every one is kept, with the labels read from it.

Counted from the record of the October 2026 Edition. Every index stays free to read, with no account; a license is for the catalog as data.

What is in it

Three tables, keyed to one another, delivered as files (JSON, CSV or SQLite) and updated with each edition.

Categories

  • Name, the function it belongs to and the index
  • How a buyer asks for it, in two phrasings
  • The landscape and analyst references it maps to
  • The date it was added, and every edition it was asked in

Products

  • One canonical name, its company and its website
  • Every spelling the models used for it
  • The categories the models name it in
  • A description, with its basis recorded
  • What its own site states: positioning, target buyer, pricing, trial, certifications, hosting

Each month

  • First-choice share and rank by category and buyer size
  • Movement against the previous edition, read against a measured noise floor
  • The sites the models cited
  • Every label traced to the answer and the sentence it came from

Beyond the answers

Beside what the models say, the catalog holds what the vendors themselves do: what they publish, where they stand in search and what they advertise, read for the products the models name.

Content

14,317,608

pages on 8,987 vendor sites, each sorted into one of 17 kinds (the table below)

8,276,584

pieces of editorial content: blog posts, news, reports, case studies, comparisons, glossaries, templates, webinars, podcasts and events; 4,526,817 carry a publication date, so how often a vendor publishes, and on what, can be counted

1,681,592

product, pricing and company pages, beside 2,719,922 documentation pages and 883,026 integration pages

2,508

upcoming events found on 641 vendors' own sites

Search

3,580

vendor sites with their search footprint: ranked keywords, top-three positions, estimated organic visits, linking sites

5,674

product and company names with their monthly searches on Google and an estimate of their demand in AI search

363

categories with Google's top ten for the searches a buyer types, set beside the sites the models cite

Paid

147,105+

Google ads counted for 2,789 advertising vendors, with their formats and when they ran

3,580

vendor sites checked for ad activity, advertising or not

928

vendor sites with an estimate of their paid search keywords and spend

The 17 kinds of page
KindPages
Blog posts4,332,403
Documentation2,719,922
Integration pages883,026
News and press853,817
Product pages772,942
Listing and index pages756,484
Company pages731,309
Reports and ebooks632,771
Glossaries and explainers530,606
Comparisons516,783
Events497,734
Templates and tools345,918
Case studies239,848
Webinars181,379
Podcasts and video145,325
Home pages138,801
Pricing pages38,540

Search, traffic and spend figures are US estimates. A vendor with no figures was not seen in that source, which is not the same as having none.

By index

IndexFunctionsCategoriesProductsAnswers, October 2026
GTM AI IndexMarketing, Sales, Customer, Revenue operations, GTM data and infrastructure, Partner and channel1126,83428,224
IT AI IndexSecurity operations, IT operations and endpoint, Cloud and infrastructure, Developer platform, Data platform, Identity and access, Network and edge1367,64934,272
HR AI IndexCore HR and payroll, Talent acquisition, Onboarding and employee experience, Performance and talent management, Learning and development, Compensation and total rewards, Workforce planning and analytics583,24514,616
Finance AI IndexAccounting and close, Planning and analysis, Spend and procurement, Receivables and billing, Treasury and cash, Tax and compliance, Equity and corporate573,16914,364

How it is kept

  1. Every month each index asks every category six ways, for a small, a mid-market and an enterprise buyer, of the same fourteen models on the same day, and a disclosed judge labels every product named.
  2. The names the models write are read to products through a versioned vendor table. A rename, a merger or a correction takes a new version, every edition is re-read under it, and the change log names it.
  3. Products are tied to their own websites and companies through one entity layer for all four indexes.
  4. Descriptions are written from what each product's own site states, and from the answers that name it where no site has been read; each records which.
  5. The noise between two identical runs is measured by a repeat and published, so a change is reported only when it is larger than chance. The method is public: gtm-ai-index.com/methodology.

Ask about a license

A license covers using the catalog inside a product, a marketplace or a directory, in internal tools and analysis, or in research, and comes with each month's update. Terms are agreed per use. Say what it is for and a reply follows with the scope, the format and the price.