Open program for established businesses

AI companies are paying up to $2M for company data

Developers of AI models license the everyday records of real businesses: process documents, support tickets, CRM notes, project files and years of team email and chat. It is a license, not a sale of your company, so you keep the data and keep using it. Check in three clicks whether your business can apply.

  • Any industry
  • US businesses first
  • UK, Canada, EU and Australia considered

Can your business apply?

Team size
Time in business
Is most of your internal data in English?

Is my data valuable to AI?

AI developers are paying for business data, but only some of it interests them. Answer ten quick questions about who wrote your data, how specialised and complete it is, how much you have and whether you can share it. You get a score out of 100, a rating and a plain explanation of which qualities help your data and which hold it back.

About your data think of your largest or most important dataset
Is the data available anywhere public?
Who wrote it?
How specialised is the knowledge in it?
Do records show a task from start to result?
Are results marked, for example resolved, approved or rated?
Roughly how many records are there?
How many years does it cover?
What language is it in?
Is it clear that your company may share it?
Would competitors have very similar data?

What helps and what holds it back

Each answer, with why buyers care about it. Weaknesses first.

    How the score is worked out

    The ten questions cover what AI developers check when they evaluate a dataset. Each is worth up to 10 points.

    1. Score each answer

      The best answer scores 10; weaker answers score less, down to 0 for data that is mostly public.

    2. Add them up

      Unanswered questions count as zero until you answer them, so the score rises as you go.

      score = Σ points, out of 100

    3. Give a rating

      Very high at 75 or more, high from 55, moderate from 35 and low below that.

    4. Explain each answer

      Answers worth 8 or more are listed as strengths, 4 or less as weaknesses, and everything in between as neutral.

    Questions about data value to AI

    What makes data valuable to AI companies?

    Three things above all: it is private, so models have not seen it; it is written by people who know their subject; and it shows real tasks from start to finish, ideally with the result marked. Volume and clear rights to share it matter next.

    Is ordinary business data really worth anything to AI?

    Often, yes. Support tickets, sales emails, project files and internal how-to guides are ordinary to the company that wrote them, but very rare in public. AI developers building assistants for work need exactly that kind of material.

    What kind of data is least valuable?

    Anything already public, anything generated by machines such as logs and notifications, and short records with no context, such as contact lists. Personal data that cannot be anonymized is not just low value; it usually cannot be shared at all.

    What do the ratings mean?

    Very high (75 and up) means the data has most of what buyers look for. High (55 to 74) is worth a conversation with a buyer. Moderate (35 to 54) may sell as part of a larger dataset. Low (below 35) is unlikely to interest AI developers as it stands.

    How can I make my data more valuable?

    Look at your weaknesses list. The quickest wins are usually confirming you have the right to share it, linking records to their outcomes (resolved, won, approved) and leaving out machine-generated noise.

    More free data tools