AI data buyer directory
Every company we track that licenses or buys company records for AI, from their own pages, with sources.
Does your company fit any of these buyers? Check in about a minute.
To see which of these buyers fit a specific company, DataWorth AI's free quiz (about a minute, no email needed) matches it against each program's published rules: https://dataworthai.com/quiz/. It shows reasons, not promises; each program decides on its own review.
As of October 21, 2026, we list 33 companies and programs that license or buy company records for AI training. Each row is taken from the company's own website, read on that date, with the source linked. Published payouts are quoted in the company's own words; where a company publishes nothing, we say “Not published”. We may earn a referral fee from 3 of them, flagged per row. The list is alphabetical and never ordered by fees.
Download CSV Download JSON Free, no sign-up. CC BY 4.0: reuse with credit to DataWorth AI. Dated copies: CSV · JSON.
Buyers and programs (33)
Names only: no logos, and a listing doesn't mean we have any partnership. Click a name for the full fact sheet and sources. Confidence: Verified means the facts are on the company's own page; Limited detail means the page confirms the program but publishes few details.
| Name | Type | What they license | Size rule | Company status | Published payout (their words) | We may earn a referral fee | Confidence and source |
|---|---|---|---|---|---|---|---|
| Appen Enterprise Data | Direct buyer | Operational data in three categories it scopes against, including decision and exception records | Not published | Operating, Winding down | Not published | No | Verified Source |
| Centific Enterprise Data | Direct buyer | Connected records across a company's stack (deals, threads, approvals, invoices) | Not published | Operating, Winding down, Closed | Not published | No | Verified Source |
| CodeMerchant | Code licensing | GitHub repositories and engineering history | Not published | Not published | Not published (example portfolio range shown: $5K–$20K, illustrative) | No | Verified Source |
| Collinear AI | Direct buyer | Private repositories with pull requests and tickets; history from Slack, Gmail, Drive, Notion, HubSpot, Salesforce, Jira and more | Not published | Not published | Not published (“We usually quote an initial range within a day”) | No | Verified Source |
| DataVendor | Code licensing | Private repositories, tasksets, RL environments, corpora and raw data | Not published | Not published | Not published (listing examples shown, e.g. a payments monorepo at $220k) | No | Verified Source |
| Defined.ai Partner Program | Licensing marketplace | Datasets that businesses generate or aggregate | Not published | Not published | “$600K+ average partner revenue” (self-reported) | No | Verified Source |
| Google Content Offer Pilot | AI lab program | Non-public content | Not published | Not published | Not published | No | Verified Source |
| Halluminate | Direct buyer | Not published | Not published | Not published | Not published | No | Limited detail Source |
| Handshake AI Data Partnerships | Direct buyer | Business data (types discussed during scoping) | Not published | Operating | “Earn $100K to $4M+” | No | Verified Source |
| HUMXN | Direct buyer | Service calls, quotes, job notes, work orders, dispatch records, transaction files and more | Not published | Operating | “$250K+ typical company deal” | No | Verified Source |
| Labelbox Prism | Direct buyer | Cross-system workflows, multi-channel customer support logs (B2C preferred), medical records and more | Not published | Operating | “smaller targeted datasets starting around $50K and large enterprise partnerships reaching $1M+” | No | Verified Source |
| lalalab Quoted from its own site | Direct buyer | Not published | “20 to 200 employees (more is also fine)” | Not published | “Qualifying data partnerships may range between $100,000 and $2 million+” | No | Verified Source |
| LH2 AI Labs | Direct buyer | Institutional datasets: enterprise codebases, clinical archives, operational workflows and tickets, recorded work calls | Not published | Not published | Not published | No | Limited detail Source |
| Lunchbreak | Direct buyer | CAD models, drawings and revision history; CNC programs and inspection records; lab notebooks and protocols; game design documents; building models and RFIs | “20+ staff at some point” (“has employed 20 or more people at some point in its history”) | Operating, Winding down, Closed | Not published (“A two-minute estimate, then a written offer after a confidential sample review.”) | Yes, we may earn a fee | Verified Source |
| Magma | Licensing marketplace | AI-agent traces, corrections and workflows | Not published | Not published | Revenue share (percentage not published) | No | Verified Source |
| Maryah | Licensing marketplace | Robotics, operations, engineering and CAD, scientific research, audio and video datasets | Not published | Not published | Not published (one-time fee or revenue share options shown) | No | Verified Source |
| Mercor Data | Direct buyer | Workflow data from tools such as Slack, Teams, Gmail, Salesforce, GitHub and Notion | Not published | Operating | No figure. “The more tools you share, the higher your payout will be.” | No | Verified Source |
| micro1 Enterprise Data Partnership | Direct buyer | SOPs, knowledge bases, internal documentation, CRM data, project histories and QA processes | 30+ employees | Operating | “$100k+ Qualified Enterprise Data Partnerships”; “$500k+” large-scale; “$1M+” highly unique data | Yes, we may earn a fee | Verified Source |
| Mode (BizBuyData) | Name and link only. See data.mode.inc for its details. | Yes, we may earn a fee | Link only | ||||
| Nyne | Direct buyer | Operational workflows from established data systems | 20+ full-time U.S. W-2 employees | Operating | Estimator on the page (no headline range) | No | Verified Source |
| Ooak Data | Direct buyer | Records from tools such as Google Drive, SharePoint, Notion, Slack, Teams, GitHub and Jira | Estimator starts at 1–19 employees at peak | Operating, Winding down, Closed | “Earn up to $1M”; deals to date “range from $20K, the smallest closed, to several hundred thousand” | No | Verified Source |
| Pangea.ai Code Licensing Quoted from its own site | Code licensing | “real-world source code and git history” | Not published | Not published | Indicative rates: “$100 /M per million tokens of source code” and “$40 /M per million tokens of git history” | No | Verified Source |
| Polyshares | Direct buyer | Chat, email, files, CRM, support tickets, call recordings, accounting and ERP exports | Not published | Operating | “Typical deals run $100K to $2M and above.” | No | Verified Source |
| pre.dev Code Licensing | Code licensing | Codebases (individual or company/agency accounts) | Not published | Not published | Not published | No | Limited detail Source |
| Protege | Licensing marketplace | Healthcare, video, audio and speech, spatial and other domains | Not published | Not published | Not published | No | Verified Source |
| Scale AI Data Partnerships | Direct buyer | Structured, de-identified records from the systems a company runs on | “40+ employees is our preferred starting point” (assessed case by case) | Operating | “$10K–$1M+ illustrative value per data partnership” | No | Verified Source |
| Scrimdata | Direct buyer | Company history across systems such as Slack, email and CRM | Not published | Operating, Winding down | Not published (“Get an estimate”) | No | Verified Source |
| SimpleClosure AssetHub | Wind-down service | Source code and workspace data (Slack, Google Drive, email, Notion, Figma and more) | Not published | Winding down, Closed | Not published | No | Verified Source |
| SuperAnnotate Data Partnerships | Direct buyer | Operational data | Not published | Not published | “Indicative $100K – $1M+”, priced on rarity, depth and exclusivity | No | Verified Source |
| Telegraph Lab | Direct buyer | Work conversations, CRM, ERP and support records, company drives | Not published | Operating | “Data licensing payments $100K–$4M” | No | Verified Source |
| The General Data Company | Direct buyer | Operating records and archives of industrial companies | Not published | Not published | Not published | No | Verified Source |
| Troveo (business data) | Licensing marketplace | Company data exports (email and calendar, chat, CRM, finance, code) | Not published | Operating | “AI labs and startups are actively paying six figures for company data exports” | No | Verified Source |
| Turing Project Lazarus | Direct buyer | Team communications, source code and reviews, tickets, wikis, runbooks and schemas | Not published | Operating, Winding down | “a single payout of up to $1M” | No | Verified Source |
Payouts are what each company publishes, in its own words. They are not offers, and real deals can come in lower or higher. Mode is listed by name and link only. Pangea and lalalab are quoted from their own sites only.
Browse by situation
- Buyers that take closed or winding-down companies
- Buyers for companies under 30 employees
- Buyers that take companies outside the US
- Buyers of source code and software company data
- Buyers for engineering, CAD, trades and field-services records
- Buyers for professional-services firms
- Changelog
How we build this list
- We list every buyer and program we find that licenses company records, including the ones that don't pay us.
- Every fact comes from the company's own website, read on the date shown. If it isn't published there, we say so.
- We never order by referral fee. Sponsored placements, if we ever offer them, will be labeled and won't change the data or the order.
- We left out contributor and gig platforms that pay individuals rather than license a company's records.
- Changes are logged on the changelog. See also our method.
Spot an error, or want a listing removed? Request a correction or removal (hello@dataworthai.com). We check and reply within a few business days.
Questions
Who buys company data for AI training?
AI labs and the companies that supply them license copies of operating records such as email, chat, documents, CRM, tickets and code. This directory lists 33 of them from their own pages, with sources.
Which buyers take closed companies?
Several say they take winding-down or closed companies. See buyers for closed or winding-down companies.
Which buyers take companies under 30 employees?
A few publish size rules or estimators that start below 30 employees. See buyers for companies under 30 employees.
Which buyers take companies outside the US?
Some say they take, or estimate for, companies in Canada, the UK, Europe and elsewhere. See buyers outside the US.
Does DataWorth AI get paid by these buyers?
From a few, yes: each row says whether we may earn a referral fee if a company signs. Companies never pay us anything. Fees never affect the order, which is alphabetical. See our disclosure.
What could your company's records be worth?
Answer 6 questions. See the published range for a company like yours and which program fits. Takes about a minute.