For software companies
Get paid for what your engineering team already knows.
Your repositories hold more than code. Commits, reviews, issues and incident write-ups record how your engineers made decisions and fixed real problems. AI labs pay for that kind of record. We license selected repositories and records on your behalf, keep secrets and customer data out and negotiate the best available price.
Systems we often see
- GitHub
- GitLab
- Bitbucket
- Jira
- Linear
- Confluence
- Notion
- Sentry
- PagerDuty
- Slack
Partner estimate for qualifying records: $100K to $1M+. Buyer interest and value are not guaranteed.
- No files or logins to start
- You pay nothing unless a license is signed
01 Why AI labs care
What your records show that AI can’t do yet.
AI can write a function. It still struggles with the work around the code: understanding a large, aging codebase, reviewing a risky change, tracing a production incident to its cause and deciding what to refactor and what to leave alone. Your history shows those decisions, with the tests and outcomes attached.
What makes software companies’ records valuable
- Long commit histories that show how a real codebase evolved, not just its current state
- Pull request reviews where engineers questioned, improved and approved changes
- Issues and incidents linked to the commits that fixed them
- Private production code that isn't already public
02 Workflows
The work with the most value.
Buyers look for complete pieces of work, start to finish. In your industry, these are the strongest.
Code review
Pull requests, review comments and the revisions made before a change was merged.
Bugs and fixes
Issue reports, reproduction steps, debugging discussion and the commit that resolved each one.
Incidents and post-mortems
Alerts, timelines, root cause analysis and the follow-up work that prevented a repeat.
Architecture and design decisions
Design docs, RFCs and decision records, including the options that were rejected and why.
Migrations and refactors
Large changes planned, staged and rolled out, with the problems found along the way.
03 Connected records
One piece of work, followed end to end.
This is what connected history looks like, and why it’s worth more than any single file.
Illustrative exampleInvented details. Not client data.
A production bug, from alert to merged fix
- AlertCheckout errors jump to 3% within an hour of the morning deploy.
- Incident channelEngineers roll back, then trace the errors to a time zone bug in a new discount rule.
- Issue trackerIssue 2291 records the cause, the customer impact and a failing test that reproduces it.
- Pull requestThe fix and a regression test go to review. A reviewer asks for a second edge case, which is added.
- Post-mortemThe write-up adds a pre-deploy check for date logic, merged the same week.
Each artifact is routine. Together they show engineering judgment under pressure, from symptom to lasting fix. Secrets, customer data and identifying details are removed before anything is shared.
04 Privacy and rights
What stays out, and what’s reviewed first.
Every industry has its own sensitive material and its own rules. For software companies, these come first.
Always excluded or removed
Before anything is shared, these stay out entirely or are replaced.
- Secrets anywhere in code or history: API keys, tokens, passwords and private keys
- Customer data in fixtures, logs, database dumps and support tickets
- Code your customers own, such as custom development delivered under contract
- Third-party and open-source code whose license terms don't allow it
- Export-controlled technology or source code
- Employee and contributor names and emails (replaced with consistent stand-ins)
Rights reviewed before any records are shared
Open-source and third-party code
Repositories often include open-source and licensed third-party code, each with its own license terms. Those terms are reviewed, and code you don't have the right to license stays out.
Customer contracts
Customer agreements may give customers ownership of custom work or restrict how their data and confidential information can be used. They're reviewed before any repository is considered.
Who wrote the code
Code written by contractors or acquired from another company is only yours to license if the IP was properly assigned. Ownership is checked before anything is included.
This isn’t legal advice. We flag what we find and encourage you to have your own counsel review. How records are protected
05 How it works
From first conversation to signed license.
We negotiate on your behalf and manage the process. You approve every term, and you pay nothing unless a license is signed.
- 01
Start with a conversation
Tell us about your business, the systems you use and how far back your records go.
No files or logins needed - 02
Agree on what’s in scope
Together we decide what’s included and what stays out. Rights and privacy are reviewed before anything is shared.
You set the boundaries - 03
Bring offers to the table
We present your opportunity to AI labs, compare the offers and negotiate price and terms.
We negotiate for you - 04
Approve and get paid
Nothing moves until you sign. Records are prepared and delivered under the terms you agreed.
Your signature, your call
06 Questions
Questions from software companies.
New to this? Start with our plain-English guide to licensing business data.
Can we license code without exposing our product?
Often, yes. You choose which repositories and time periods are in scope, and you can exclude current or sensitive work. The license sets how buyers may use the code, and secrets and customer data are removed first.
Which engineering records are most valuable?
Long, active histories: years of commits with the reviews, issues and incidents linked to them. Private production code with real-world complexity is worth more than small or public projects.
Do you need access to our GitHub organization?
Not to start. A conversation is all it takes. If you decide to go ahead, the scope you approve decides exactly which repositories are prepared, and nothing is shared until you sign.
