How we rate AI tools

What the band means

Each tool shows one band: Higher, Medium, Lower, or Unrated. We publish a band, not a precise number, because a band plus our reasons and sources is a fair, defensible assessment, while a single figure on a named product is a claim a vendor could dispute. Areas we cannot verify are left Unrated and pulled out of the calculation, so a tool is never marked down for information we could not find. If we cannot verify at least half of the four areas, the tool stays Unrated.

The four areas and their weight

Data location and compliance carry extra weight because our customers are security and compliance teams.

Dated, sourced, and correctable

Every rating carries the date we checked it and links to the public sources behind it. If you make one of these tools and think a rating is wrong, request a correction.

How we measure coverage

The trust ratings tell you how safe a tool is. The coverage numbers tell you how much of the AI landscape this catalogue can actually see. We track three things:

Reference coverage

We check the catalogue against a reference list we maintain of the most-adopted workplace AI tools, built from public adoption surveys, and report what share we detect. The list is ours, dated, and published with every run. It is not a claim that we identify every AI tool in existence. A tool “counts” if Northbeams can identify it by name in live telemetry, not merely if we know it exists.

Field identification rate

In live telemetry from connected organizations, this is the share of AI-signal events where Northbeams attributes the activity to a named tool rather than leaving it as an unknown AI call. A higher rate means fewer blind spots in the usage log.

Catalogue scale

The number of distinct AI tools this catalogue can identify by name. It grows as the daily refresh pipeline discovers new tools; it is a ceiling on coverage, not a claim that every listed tool is active in your organization.