Policy

Methodology

How records are defined, sourced, and maintained.

Scope

The index covers publicly available AI products: end-user applications, developer tools, hosted services, and open model families with a public offering. Internal or unannounced products are out of scope. The current dataset is a seed set and is not yet representative of the whole market.

Records and fields

Each product is a single JSON record in data/products/; the schema is documented on the dataset page. Records describe the product as it currently exists; historical notes belong in the change history, not the description.

Business model values

When a product mixes models (e.g. an open model with a paid hosted service), the record reflects the primary public offering.

Sources and provenance

Every record must cite at least one source, preferring official material (product pages, official announcements). Sources are listed on each product profile and shipped with the data exports.

Updates and corrections

Corrections are made by editing the record's JSON file in the source repository — open an issue or a pull request. A record's updated field changes only when the record itself is reviewed or modified.

Versioning

Aggregate exports carry a dated version; the authoritative change log is the git history of the repository.