Who buys company data for AI training
As of October 25, 2026, we list 17 buyers: companies that license internal business records (email, chat, documents, CRM, finance, tickets or code) for AI and pay the company that holds them. We leave out brokers, listing sites and other go-betweens that never handle the data themselves. Each profile uses the same fields, drawn from the buyer's own pages.
Start here
Buyers marked "We may be paid" pay us a referral fee if a company we sent signs; "Featured" buyers are ones we highlight for their scale and backing. You pay nothing either way, and every buyer below is described on the same fields. How we make money.
Handshake AI We may be paid
20+ employees · published range $100K–$4MHandshake AI licenses de-identified operational data from businesses and prepares it for frontier AI labs to train and evaluate models.
HUMXN We may be paid
published range $250K+HUMXN, which opened publicly in October 2026, reviews rights, de-identifies and formats company records, licenses them to approved AI buyers as the licensor, and pays the company per license.
Mercor We may be paid
published range $20K–$1.5MMercor connects to a company's work tools, automatically extracts and de-identifies the workflow data, licenses it to AI labs, and pays the company.
micro1 We may be paid Featured
30+ employees · published range $100K–$2Mmicro1 pays established companies, and companies that have closed, for operational documentation and workflow data that it prepares for frontier AI development.
Mode We may be paid Featured
50+ employees · published range $100K–$5MMode buys an agreed, de-identified copy of a company's internal operating records for AI research and training, with no broker in between.
The General Data Company We may be paid
no minimum or range publishedThe General Data Company licenses anonymized operating records, mainly from industrial and manufacturing firms, and supplies them to AI labs and manufacturers.
Troveo We may be paid
no minimum or range publishedTroveo prepares an anonymized copy of a company's internal records and licenses it to AI labs and AI companies on the company's behalf, sharing the revenue from each sale.
One-time payment (7 buyers)
Licenses a defined copy of your records and pays an agreed amount once, usually on delivery or acceptance.
Target: Owners who want one known amount.
Revenue share (3 buyers)
Prepares your records and pays you a share of each sale to AI companies, with no guaranteed total.
Target: Owners comfortable trading certainty for repeat payments.
Per repository (1 buyers)
Pays a published rate per code repository or codebase.
Target: Companies that built their own software.
Priced after scoping (6 buyers)
Publishes no figure. You get a number only after it has reviewed your records, then terms in writing.
Target: Owners ready to compare written offers.
Pays a one-time amount for your records
Mercor We may be paid
published range $20K–$1.5MMercor connects to a company's work tools, automatically extracts and de-identifies the workflow data, licenses it to AI labs, and pays the company.
micro1 We may be paid Featured
30+ employees · published range $100K–$2Mmicro1 pays established companies, and companies that have closed, for operational documentation and workflow data that it prepares for frontier AI development.
Mode We may be paid Featured
50+ employees · published range $100K–$5MMode buys an agreed, de-identified copy of a company's internal operating records for AI research and training, with no broker in between.
Nyne
20+ employees · published range $100K+Nyne licenses de-identified operational data from established companies and prepares it as training and evaluation data for frontier AI labs.
Polyshares We may be paid
published range $100K–$2MPerpetual exclusive licenseTrack record: four unnamed deals, self-reportedPolyshares licenses operational records directly from a company, pays the company itself, and then anonymizes the data and supplies it to AI labs.
Telegraph Lab
published range $100K–$4MTelegraph Lab pays businesses to license a prepared copy of their existing records for AI training and evaluation, under a written agreement that names who may use the data.
Turing (Project Lazarus)
published range up to $1MTuring's Project Lazarus pays a single amount for a company's operating record (team communication, code and engineering history, wikis and runbooks) to train frontier AI models.
Pays you a share of each sale of your records
HUMXN We may be paid
published range $250K+HUMXN, which opened publicly in October 2026, reviews rights, de-identifies and formats company records, licenses them to approved AI buyers as the licensor, and pays the company per license.
Troveo We may be paid
no minimum or range publishedTroveo prepares an anonymized copy of a company's internal records and licenses it to AI labs and AI companies on the company's behalf, sharing the revenue from each sale.
Defined.ai
no minimum or range publishedDefined.ai, a long-standing AI training-data company, reviews a partner's datasets for quality and rights, licenses them to AI buyers under its own license, and pays the partner licensing fees from each sale.
Pays per repository (software code)
pre.dev
published range $740–$6KA program that values private Git repositories and licenses chosen ones, non-exclusively, to AI labs training coding models.
Priced after scoping: no figure until your records are reviewed
Handshake AI We may be paid
20+ employees · published range $100K–$4MHandshake AI licenses de-identified operational data from businesses and prepares it for frontier AI labs to train and evaluate models.
The General Data Company We may be paid
no minimum or range publishedThe General Data Company licenses anonymized operating records, mainly from industrial and manufacturing firms, and supplies them to AI labs and manufacturers.
Halluminate
no minimum or range publishedA Y Combinator-backed research lab that builds AI training environments for finance and consulting work and invites firms to discuss licensing their data and workflows.
LH2 AI Labs
no minimum or range publishedA San Francisco data company that licenses institutional data from hospitals, enterprises and factories and packages it for AI labs.
Ooak Data
published range $20K–$1MOoak Data, a Y Combinator-backed AI research lab, pays companies (including closed ones) to license their anonymized work-tool data for AI training.
Scrimdata
no minimum or range publishedScrimdata licenses a company's connected operating history across its work tools, anonymizes it, and turns it into training environments for AI labs.
All programs at a glance
| Buyer | How it pays | Min. size | Accepts | Published range | Checked |
|---|---|---|---|---|---|
| Defined.ai | Revenue share | – | operating, closing, closed | – | 2026-10-25 |
| Halluminate | Priced after scoping | – | operating | – | 2026-10-25 |
| Handshake AI | Priced after scoping | 20+ | operating | $100K–$4M | 2026-10-25 |
| HUMXN | Revenue share | – | operating | $250K+ | 2026-10-25 |
| LH2 AI Labs | Priced after scoping | – | operating | – | 2026-10-25 |
| Mercor | One-time payment | – | operating | $20K–$1.5M | 2026-10-25 |
| micro1 | One-time payment | 30+ | operating, closing, closed | $100K–$2M | 2026-10-25 |
| Mode | One-time payment | 50+ | operating | $100K–$5M | 2026-10-25 |
| Nyne | One-time payment | 20+ | operating | $100K+ | 2026-10-25 |
| Ooak Data | Priced after scoping | – | operating, closing, closed | $20K–$1M | 2026-10-25 |
| Polyshares | One-time payment | – | operating | $100K–$2M | 2026-10-25 |
| pre.dev | Per repository | – | operating, closing, closed | $740–$6K | 2026-10-25 |
| Scrimdata | Priced after scoping | – | operating, closing | – | 2026-10-25 |
| Telegraph Lab | One-time payment | – | operating | $100K–$4M | 2026-10-25 |
| The General Data Company | Priced after scoping | – | operating, closed | – | 2026-10-25 |
| Troveo | Revenue share | – | operating | – | 2026-10-25 |
| Turing (Project Lazarus) | One-time payment | – | operating, closing, closed | up to $1M | 2026-10-25 |
Ranges are each buyer's own published figures, not offers. A dash means none is published. Grouped by how each buyer pays; within each group, buyers that pay us or are featured come first, then alphabetical. How we research and update this list.