Repositories with full history
Commits, branches and merges for code you own, with secrets scanned out.
For software companies & game studios
Short answer: Software companies and game studios can license code they own outright, including commit and pull-request history, code review threads, tickets, design docs, incident postmortems and, for studios, original 3D assets and level data, to AI developers. Open-source and third-party code, licensed engine or marketplace assets, customer data and secrets stay out.
AI developers training coding and game-building agents want more than code snapshots: the pull requests, review comments, tickets and postmortems that show how a team ships. We help companies scope what they own outright, keep third-party code, licensed assets and secrets out, and reach the buyers asking for it.
Last updated · Sources linked inline · Buyer figures as published, observed Oct 16, 2026 (micro1, Mercor and GDC re-checked Nov 3, 2026)
Free to start. No files or logins needed.
In four steps, with no cost to start: share the basics, scope what the firm owns, get matched to buyers whose published criteria fit, then sign directly with the buyer, who pays you.
Headcount, years of records and the systems you use. No files, no logins, no client names.
We separate code and assets you own outright from open-source, vendored and licensed material, and check customer and publisher agreements before anything is scoped.
Only programs whose published criteria fit a software company or game studio of your size, location and record language.
The buyer handles de-identification under a written agreement and pays your firm directly. You keep your originals.
The engineering records AI buyers ask software teams for: The process around the code is often worth as much as the code: who proposed a change, how it was reviewed, and what broke.
Commits, branches and merges for code you own, with secrets scanned out.
Review comments and revisions show engineering judgment.
How work is scoped, estimated and closed.
Alerts, debugging threads and written postmortems, de-identified.
Original 3D models, animations, levels and design docs your studio owns outright.
Buyer programs that list software, code or gaming records. Public terms only. Size, location and language rules differ by program, so eligibility is confirmed buyer by buyer.
| Buyer | What they publish | Who they say qualifies | Source |
|---|---|---|---|
| Mercor | No amounts published; says it charges no fees or costs and pays by wire when the anonymized dataset is delivered, typically 2–4 weeks after signing. Payout rises with the number of tools connected. As published · verify | Target industries include legal, healthcare, finance and insurance, professional services, manufacturing, retail, real estate and logistics. Its CEO's Oct 23, 2026 post targets companies with 100+ employees and data in code, gaming, manufacturing, healthcare or professional services. | Buyer page checked directly Nov 3, 2026 CEO post, Oct 23, 2026 |
| micro1 | Says it pays $100K to $3M+ (tiers: $100K–$500K, $500K–$3M, and $3M+ for highly unique data). Payment within 30 days of invoicing, after a 30-day review of the delivered data. As published · verify | 30+ full-time employees, mature operations, records primarily in English. Prioritizes US companies, then other Western markets; says current demand is strongest for companies based in the US, UK and Canada. Licenses only to US and allied-country labs and enterprises. | Buyer page checked directly Nov 3, 2026 |
| Collinear AI | No amounts published; initial range quoted within a day. As published · verify | Code plus business-app history (e.g., pull requests linked to tickets). Price rises with 15+ peak staff and several tools. | Buyer page via Who Sells Data register, observed Oct 16, 2026 |
| Turing (Project Lazarus) | A single payment of up to $1M, confirmed after sampling (usually within a week). As published · verify | Operating, winding-down or closed companies. Wants communications, tickets, wikis and databases; code alone doesn't qualify. | Buyer page via Who Sells Data register, observed Oct 16, 2026 |
Amounts are each buyer's own published ranges, ceilings or tiers, not offers, and they change. Most buyers are US companies and don't always state a currency, so confirm it in writing. Size, location and language rules differ by buyer. Buyer names describe their public programs only. How the process works.
The line for software companies and game studios
Only what you own outright, and never secrets. For software, the risks are licence contamination and security, as well as privacy.
Sources: GNU: GPL FAQ · Unity Asset Store terms · GitHub: about secret scanning · Mercor CEO post, Oct 23, 2026 · micro1: data partnerships
General information, not legal advice. The rules depend on where your firm and its clients are; we ask every firm to confirm scope with its own counsel or regulator before signing. Jurisdiction notes: privacy & compliance hub.
An illustrative case: a 45-person B2B software company with 8 years of history in GitHub, Jira and Confluence. Vendored and open-source code is excluded, secrets are scanned out and customer data never leaves. The licensable set is first-party repositories with pull-request and review history, tickets and design docs. Buyers' published tiers put a company this size around $100K–$500K per licence (micro1's first published tier); micro1 lists software and engineering among its categories, and Mercor's CEO names code and gaming among the data it wants (from 100+ employees).
Illustrative only. Bands come from buyers' published tiers; not an offer or a valuation.
Start with the pillar guide: how to license your firm's records to AI.
Three answers give you a rough range based on what buyers publish. The rest helps a buyer take you seriously: every question after step 1 is optional. No files, logins or client names.