What data do AI labs want from UK companies?
Records of real work. AI labs want multi-year, English-language operational data that shows how a business actually runs: conversations, documents, decisions and the systems that tie them together. They use it to train and evaluate AI models and agents on real business workflows, so complete, linked records from established companies are worth the most.
Which types of company records are most valuable?
The most valuable records show a task from start to finish. A thread where a request comes in, is discussed, approved and recorded in a system is worth more than isolated files. Communications, documents and system records are each useful, and they are worth most when they link to each other across several years.
| Communications | Teams, Slack and email threads, with roles and timestamps preserved |
|---|---|
| Documents and procedures | SOPs, tenders, method statements, contracts, reports and version history |
| Systems and records | CRM history, support tickets, project tasks and finance workflows |
| Cross-system links | Where a conversation, document and system record relate to the same piece of work |
What makes a UK company a good fit?
Demand is strongest for companies with 30 to 500 employees, at least three years of English-language digital records and mainstream tools such as Microsoft 365, Google Workspace, Slack, a CRM or a ticketing system. That size gives enough real collaboration to be useful while keeping approvals simple. If you’re close, it’s still worth asking.
| Headcount | 30 to 500 employees |
|---|---|
| History | Three or more years of digital records |
| Language | English |
| Tools | Microsoft 365, Google Workspace, Slack, a CRM or a ticketing system |
| Location | UK companies |
Which sectors are AI labs most interested in?
Demand is strongest in sectors with rich, process-driven work: accounting and finance, legal and contracts, construction and engineering, software and engineering, customer support and sales, and operations and logistics. Other sectors can qualify too, as long as the records show how work is actually done.
| Accounting and finance | Month-end, VAT returns, reconciliations, approvals and reporting |
|---|---|
| Legal and contracts | Conveyancing, contract review, due diligence and compliance procedures |
| Construction and engineering | Tenders, RAMS, valuations, variations and site reporting |
| Software and engineering | Code history, incident reviews, product planning and documentation |
| Customer support and sales | Ticket workflows, CRM history, playbooks and QA |
| Operations and logistics | SOPs, scheduling, fulfilment, procurement and supplier management |
Why do AI labs want real business data?
Because AI agents are increasingly built to do business tasks, and they need examples of how those tasks are really done. Real records show the steps, judgement calls, handovers and exceptions that public web text rarely captures. Labs also need documented provenance, so their legal teams can show the data was obtained and licensed properly.
AI labs can read about licensing rights-cleared UK data from us.
What data won’t AI labs take?
Some data is excluded from every licence, whatever the buyer offers: HR and payroll files, special category data such as health information, payment and bank details, passwords and credentials, and client material the company doesn’t have the right to license. Personal identifiers are removed or replaced before any data is shared.
See what’s never included and how de-identification works.
How do you find out if your company’s data qualifies?
Ask for a revenue estimate. Tell us your sector, headcount, years of records and the systems you use, and we’ll send a private range with no data shared and no obligation. As of October 2026, most UK companies we assess fall between £100,000 and £500,000.
Get a revenue estimate, or read how company data is valued.
Common questions
Do AI labs want our customers’ personal data?
No. They want records of how work is done. Names, contact details and other identifiers are removed or replaced before anything is shared.
Is a small company’s data useful?
Demand is strongest from 30 employees upwards. Smaller companies can still ask, especially if they have long, complete records in a sector in demand.
Do we need all of these systems?
No. Most companies we assess use some combination of Microsoft 365 or Google Workspace, Slack or Teams, and a CRM or ticketing system.
Who are the AI labs?
Vetted AI developers, many of them US-based, that train and evaluate AI models and agents. Every buyer is vetted and only sees a manifest and redacted sample before a licence is signed.