The plain answer

Why do AI companies want to know how my business works?

Updated

Because AI models are now being trained to do the work inside companies, and the best teacher is a real firm's record of how a request turns into a result. Some buyers license those records from the business; your staff's personal details have to come out first, and UK data law decides what can go.

Ask about my process data

What the buyers say they are after

Mercor puts it in one line: frontier AI models are increasingly being trained to work inside companies, navigating real tools, workflows and decisions. Maryah says linked records show how real work gets done, from a customer's first email to the final outcome.

Source: Who buys workflow traces and operating records

Grepped describes the same thing as tickets and project work captured end to end: the brief, the steps taken, the back-and-forth and the finished result. Troveo says buyers pay a premium for work they can follow across communication channels, knowledge and code repositories, and operational tools.

Source: Who buys workflow traces and operating records

Which firms they look for

micro1's referral page asks for established companies with 30 or more full-time employees, mature business operations and substantial historical operational data. Nyne asks for three or more years of real operational workflows that teams already run on. None of the twelve buyers on our board publishes a price list; they quote after seeing what a firm holds.

Source: Who buys workflow traces and operating records

What UK law asks before any of it leaves

A firm's records name its staff and customers, so passing them on is processing under the UK GDPR, and Article 6 allows processing only on one of its listed lawful bases. Where processing using new technologies is likely to result in a high risk to people's rights, Article 35 requires an impact assessment before it starts.

Source: Process data employee monitoring law

The ICO's guidance for employers says a firm can only change its purpose for information gathered by monitoring workers if the new purpose is compatible with the original one, if it gets consent, or if it has a clear obligation or function set out in law.

Source: ICO, data protection and monitoring workers

What comes out before a buyer sees anything

  • Names, contact details and anything else that identifies a member of staff or a customer.
  • Anything gathered by monitoring staff, unless the new use passes the ICO's purpose test.
  • Customers' data your firm holds under their instructions, unless the customer agrees in writing.
  • Passwords, keys and other secrets that appear in exported records.

OpenAI's data partnership page says it is not seeking datasets with sensitive or personal information, or information that belongs to a third party, and Grepped says it strips customer names, contact details and identifying information before anything leaves processing.

Source: Who buys workflow traces and operating records

A note to forward to your business partner

"AI companies are licensing records of how firms like ours actually work. Ours might be worth something, once we know what the staff and customer details allow. Twenty minutes on Friday to see what we hold?"

Ask about your own records

Tell us which systems you run and how many years they hold on the enquiry form. The rules for logging staff are on employee monitoring laws UK.

Tell us about the process data your company holds

Most process-mining vendors leave ownership to the contract you signed. Tell us which tool holds your logs and how the company stands, and a person will reply.

The process data

What is most of the data?

  • A person reads every process data enquiry
  • No mining platform on the board is told you asked
  • We never ask for event logs, exports or staff names in the form

Ask what your process data could be licensed as

Most process-mining vendors leave ownership to the contract you signed. Tell us which tool holds your logs and how the company stands, and a person will reply.

Sources

This site's own datasets, every figure quoted from its source and dated:

  1. Microsoft Power Automate (Product Terms)
  2. Celonis
  3. UiPath
  4. UiPath
  5. Celonis
  6. UiPath
  7. Skan
  8. Celonis
Ask about my process data