Company data licensing
Sell your company data to AI companies
AI labs pay to license real business records as AI training data: Slack and email threads, CRM history, support tickets and internal docs. Answer a few questions and we'll tell you whether yours could qualify. It's a licence, not a sale, so the data stays yours.
Where the data lives
The data AI labs pay for is already in your tools.
Slack threads, email, support tickets and CRM history record how your business actually makes decisions. Years of those records are what labs license for training. If your team has used tools like these for a few years, it's worth checking what you hold.
Data licensing, explained
What is data licensing?
Data licensing is when a company gives an AI company the right to use copies of its business records, such as chat threads, support tickets, CRM history and documents, as AI training data for a set purpose and period. The company keeps ownership, and a non‑exclusive licence lets the same records earn more than once.
Paid by the companies building AI
AI labs and the data companies that supply them pay to license real business records.
Licensed, never sold
You grant the use of a copy for a set purpose and period. The original stays yours.
Only what you own
Only records your company has the rights to are licensed. Data you hold for clients stays out of it.
Answer a few questions and we'll tell you whether yours could qualify.
Check your dataWhy now
Why AI companies are buying business data now
AI models have already trained on most of the public internet. What they still lack is a record of real work: decisions made across weeks of messages, approvals and fixes. That record only exists inside companies, so AI companies are now paying to license it.
One AI data company
One airline's records
One healthcare dataset
What labs actually buy
What AI companies look for in business data
Size alone doesn't decide what records are worth. AI labs pay for depth: records that show how a business actually makes decisions, and what happened next.
Years of history
A long-running record shows how decisions played out over time, not just a single moment.
Outcomes attached
Closed deals, resolved tickets and paid invoices let a model connect a decision to its result.
Reasoning written down
Threads where people explain why they chose something, not only what they did.
Specialist work
Decisions from finance, legal and engineering teams are among the records buyers name most often.
What qualifies
What business data qualifies?
Buyers mostly look for companies with 30 or more people, records in English and at least two years of history, with US companies first in line. If that sounds like you and your company owns the records, it's likely to count. The free check confirms it.
Conversations
Chat and email threads where real decisions get made.
Calls and meetings
Recordings and transcripts of teams talking problems through.
Documents and knowledge
SOPs, proposals, reports and wikis: how the company says work should be done.
Sales and CRM
Deals from first touch to closed or lost, and every step in between.
Customer support
Problems captured end to end, from complaint to resolution.
Projects and engineering
Plans, issues, designs and project history, from idea to shipped.
Finance and operations
Invoices, reconciliations, approvals and orders from the systems that run the business.
Not sure your stack counts?
Tell us what you use and we'll give you a straight answer, even if it's no.
Check your dataOut of scope: data you hold on behalf of clients, and anything the company doesn't have the rights to. Got private codebases? License your code separately →
Who it's for
Which businesses can sell their data to AI companies?
Sector matters less than depth. If your team has years of records showing how work gets done, you're likely a fit.
Software and SaaS
Support queues, product specs and sprint history.
Agencies and consulting
Proposals, internal project threads and retros.
Finance and accounting
Month-end close, reconciliations and approvals.
Legal and compliance
Contract reviews, redlines and compliance calls.
Healthcare
Scheduling, billing, claims and admin workflows.
Customer support and BPO
Tickets, chats and escalations with outcomes.
E-commerce and retail
Orders, returns and customer conversations.
Logistics and supply chain
Dispatch, exceptions and carrier escalations.
Recruiting and staffing
Pipelines, screening notes and placements.
Construction and trades
Job tickets, site reports and change orders.
Gaming studios
Design docs, build histories and playtest notes.
A free check tells you whether your records could qualify.
How it works
How to sell your company data to AI companies
Four steps, and nothing is licensed until you've agreed the terms.
Check
01Tell us which tools your team uses and how many years of history they hold.
Review
02We look at what you have and tell you whether it could qualify and what affects its value.
Agree terms
03You decide which sources, date ranges and uses are included before you sign.
Get paid
04The data is licensed for a set use and term. You're paid for the licence and keep the original.
Free to check, no commitment.
Licence or sale
Licensing your data vs selling it
A sale hands the data over for one price. A licence lets the buyer use a copy on agreed terms, and you keep the original.
Key terms
The terms, in one place.
What a deal with us looks like, before you fill anything in.
- Who pays you
- AI labs and the data companies that supply them. Pangea runs the check and handles the process.
- Deal size
- Around $100K to $2M, and more for large, deep records. You get a real figure after a review.
- Structure
- A licence, not a sale. The buyer uses a copy for an agreed purpose and term.
- Cost to you
- Nothing. No fee to check and nothing to pay upfront.
- Who qualifies
- Any sector. What counts is years of records showing how work gets done.
- Scope
- You approve it before you sign: which tools, which date ranges and which uses.
- Personal details
- Removed first. Names, emails, phone and account numbers come out before the record is shared.
- Access
- Read-only, and only to the sources you approve. Nothing is changed or deleted.
- Ownership
- You keep the original and keep using it as before. A non-exclusive licence can earn again.
Free to check, no commitment.
Frequently asked questions
Everything you ask us first.
Still unsure? Start the check. Someone reads it and replies, even if the answer is that your data isn't a fit.
Check your dataHow much is my company's data worth?
It depends far more on depth than on size. A record covering many years, with outcomes attached (closed deals, resolved tickets, paid invoices) and specialist decisions in it, is worth much more than a large pile of shallow data. Company data licences in the market currently range from around 100K USD to over 2M USD for large, deep records. You only get a real number after a review.
How do I sell my company's data to AI companies?
Start with a check of what you hold: which tools your team uses and how many years of history they contain. If there's a fit, you agree the scope and terms, personal details are removed from a read-only copy, and the de-identified record is licensed to an AI lab for a set use and term. You're paid for the licence and keep the original data.
Is it legal to sell or license company data to AI companies?
Generally yes, when the company owns the data and personal information is handled lawfully. Three things decide it: whether the company owns the records, what your customer and supplier contracts allow, and how personal data is removed. All three are checked before anything is licensed. This isn't legal advice, so involve your own counsel before you sign.
Who buys company data?
AI labs building frontier models, and the data companies that supply them. They want real records of how work gets done, because that is the part of the world models haven't been able to learn from public text.
What's the difference between licensing and selling data?
With a licence you keep ownership and the buyer may only use the data for a set purpose and term. With a sale, ownership moves to the buyer. Licensing means you keep using the data exactly as before, and the record can sometimes be licensed again later.
Will our customers' or employees' details be exposed?
No. Names, email addresses, phone and account numbers and addresses are removed or replaced before the record leaves processing. Roles, timing and the work itself are kept, because that's what carries the value.
Do we need to clean up or export anything?
No. The first check only asks which tools you use and roughly how much history they hold. If you go ahead, access is read-only to the sources you approve, and extraction and de-identification are handled for you.
What does it cost?
Nothing. There is no fee to check your data and nothing to pay upfront.
How long does it take?
The check is a short form. A person reviews every submission and gets back to you. From there, timing depends on how much data you have and what buyers are looking for, so we'll give you a realistic timeline once we've reviewed yours.
Can we include data we hold for clients?
No. Only records the company owns, such as its own chat, email, CRM and files. Anything you hold on behalf of clients is out of scope.
What about our private codebases?
They can be licensed on their own, at standing per-token rates, separately from the rest of your data. See the code licensing page for how that works.