961 AI offerings on one map: decide now what you will use in 2027.

Most AI market overviews start with a vendor list. Ours starts with the question a buyer has to answer: which AI belongs in this part of the business, what does it cost, what evidence supports it, and which rules apply when? Every step of the method serves that question. Here it is in plain words.

1. Record what providers say, with a date

Each entry in the Good AI Prism is recorded from the provider's own pages: what the offering does, who it is for, where it runs, how it is priced and what it documents about data and human oversight. Every fact carries its source and the date it was read. For the first research round, provider information was reviewed on 19–20 October 2026. When a page changes, the next update records the change and keeps the earlier version on file.

2. Sort the market onto one map

The first research round holds 998 market entries, 961 of them AI offerings. Each one is placed on a single wheel: 24 topic areas, from customer service and finance to science and the public sector, with the technology layers underneath, such as models, data, compute and governance. One map means one vocabulary, so a buyer in HR and a buyer in legal can compare like with like. You can open the Prism map and see where each offering sits.

3. Read deeper where decisions are made

A map shows breadth; decisions need depth. Where buyers face real choices, the research goes further: 194 depth-research records look closely at documentation and at the evidence a provider offers for its claims. They are records about offerings, not a count of products. They feed the profiles in the Good AI Prism registry, where each card shows its facts with their dates.

4. Put the market next to the buyers

What can be bought is only half the picture. The study sets the offer side against what organisations actually do, using a separate synthesis of external studies from more than 30 economies. The figures differ because they measure different things. About nine in ten organisations in a global survey report regular AI use (McKinsey, Aug 2026); official statistics count one EU firm in five with ten or more staff (Eurostat 2025). We report both and keep them apart. In every national study we compared, large firms use AI 1.6 to 4.3 times as often as small firms.

5. Add the rules and their dates

A purchase that ignores the calendar is a risk. Each edition names the rules that apply and when they start: US state rules on automated decisions from 1 January 2027, EU duties for listed high-risk uses from 2 December 2027. The EU AI Act Navigator gives a first orientation for your own use case. The briefs go further for their field. The Health & Life Sciences brief sets out the dates for AI in medical devices, and the AI Models brief the duties for general-purpose models.

6. Publish in layers

The same evidence appears at the depth each reader needs. The Global market study covers the whole market in 41 pages with 18 exhibits; the Germany & US Focus edition reads it for the German Mittelstand and US businesses. Sector and function briefs take one field at a time. The topic area pages, such as Customer Service or HR & Recruiting, turn the map into cards you can filter and compare side by side. Every edition appears in English and German.

7. Read it again

The Good AI Prism is a longitudinal programme. Each update reads the same sources again, so readers can see how a market moves: who arrives, who changes a promise, who leaves. Providers can submit corrections; confirmed corrections appear, dated, in the next update. The rules are set out on the study methodology page.

Who does the work

ETHORITY is built as an AI-first organisation: people and AI agents work as one team in one system, and a named person is accountable for every decision. The Good AI Prism is produced this way. Sten Franke leads the research design; his first market map was a web directory for Germany in 1997. The habit has a public record: the German Social Media Prisma charted the social web from September 2009, and the Social Media Prisma paper tells that story.

What this means for you

Start with your question, not with a vendor. Find your area on the map, read the brief for your sector, open the cards, compare the offerings that fit and price the choice with the AI ownership cost calculator. Every number you take into a board paper keeps its source and its date, so the decision can be checked later. That is the point of the method: AI decisions that say what should improve, what evidence supports the choice and who is accountable for the result.