Your company produces data every day just by operating: Slack threads where a pricing decision got made, support tickets that show how an edge case was resolved, CRM notes from a two-year sales cycle, the SOPs that tell a new hire what “done” looks like. Until recently that record was a cost of doing business. Now AI labs pay for it.
This guide explains the market in plain terms for owners, founders, COOs and finance leads in the US. It covers what buyers want, what drives the price, how a deal is structured, and the questions to settle with your lawyer before anything leaves your systems.
Why AI labs pay for operational data
Large AI models were first trained on public text. The next step is teaching AI agents to do real work: triage a ticket, follow up on a deal, reconcile an invoice, hand off a project. Public internet text doesn’t show that. Your internal record does.
Labs use this data to build simulated workplaces, sometimes called “RL gyms,” where agents practice multi-step tasks, and to evaluate how well agents perform. In April 2026, Gizmodo reported (citing Forbes) that wind-down firm SimpleClosure had handled nearly 100 deals licensing startups’ Slack messages, emails and documents for this purpose.
The demand is for context and reasoning: not just what happened, but the back-and-forth that shows why.
What counts as operational data
Operational data is the working record of a business. Buyers commonly name:
- Communication: Slack or Teams workspaces, email archives
- Customer systems: CRM history (HubSpot, Salesforce), support tickets
- Knowledge: SOPs, internal wikis, knowledge bases, project histories, QA processes
- Files: Drive or SharePoint documents, spreadsheets, design files
- Recordings: sales and support call recordings
- Finance and operations: accounting and ERP exports
Read the full breakdown in What is operational data?
Not sure if your data qualifies? The intake takes a few minutes.
How a data licensing deal works
Most deals follow the same arc:
- Intake. You answer a short assessment about your systems, headcount and history.
- Review. The buyer looks at a sample of the record to judge quality and fit.
- Offer and terms. You agree the scope (which systems, which dates), anonymization rules, use limits, term and price.
- Preparation. Identifying information is stripped before anything moves.
- Delivery and payment. Data is delivered and approved, then paid. Polyshares states net 30 to 60 days.
Step-by-step detail: How data licensing works.
What your data is worth
There is no public price list, but buyers publish ranges:
| Buyer | Stated range | Source |
|---|---|---|
| micro1 | $100k+ qualified enterprise; $500k+ large-scale; $1M+ highly unique | micro1 data partnerships page |
| Polyshares | $100K to $2M+, typically | Polyshares homepage |
| Wind-down deals (per Forbes via Gizmodo) | $10,000 to $100,000 per company | Press reporting, April 2026 |
The biggest drivers are depth of history, visible reasoning (discussion, not just outcomes), domain expertise, and uniqueness. Raw volume matters less than most owners expect. See How much is company data worth?
Is it legal and safe?
Licensing your own business records is generally possible, but three questions decide what you can include:
- Do you own it? Data you hold on behalf of clients is usually off-limits.
- What do your contracts say? Customer agreements, NDAs and data processing terms may restrict use.
- Does it contain personal information? Privacy laws such as California’s CCPA/CPRA apply to personal data, which is why anonymization comes first.
This is not legal advice. Read Is selling company data legal? and How anonymization works, then take the checklist to your lawyer.
Licensing vs selling
A license is not a sale. You keep the data and the right to keep using it; the buyer gets defined rights for a defined term. That also means it is non-dilutive: no equity, no debt. More in Data licensing vs selling.
Who buys company data
You can work with a direct buyer that pays you and supplies AI labs (Polyshares describes itself this way), a data partnership program run by an AI data company (micro1), or a marketplace or broker that lists your data and takes a cut. Compare them in Who buys company data? and our reviews.
Every chapter of this guide
Frequently asked questions
Can a company really get paid for its internal data?
Yes. AI labs need realistic examples of how work gets done to train and test AI agents. Data buyers such as micro1 and Polyshares publicly state they license operational data from companies, with stated deal sizes starting around $100K for qualified records.
Do I lose ownership of my data if I license it?
Not under a license. Both micro1 and Polyshares state that the company keeps ownership of its underlying data. A license grants defined rights to defined material for a defined term.
What kind of company qualifies?
Buyers look for real operating history and records that show how decisions get made. micro1 states it targets operationally mature companies with 30+ employees, US first. Polyshares states any company with real operating history in any sector can apply.
Is it safe to share company data with an AI data buyer?
Reputable buyers strip personal and identifying information before data moves, limit access, and bound use by contract. You should still review the scope, anonymization method, retention and deletion terms with counsel before signing.
How long does a data licensing deal take?
Polyshares states an offer arrives within one to two weeks of intake and payment lands net 30 to 60 days after the data is approved and delivered. micro1 describes a four-stage process of assessment, evaluation, discovery call and partnership.