License as-is
typical range
Your records, de-identified and delivered
- Best-buyer matching across labs
- License negotiated for you
- De-identification before export
- Sample review before delivery
- Paid on lab acceptance
Typical ranges labs pay the companies that own the data. You get paid by the lab; we’re the broker in between.
50–2,000 employees · licenses project records, drawings and site imagery from finished builds.
typical range
Your records, de-identified and delivered
Stronger terms · typical range
Training-ready datasets that reach more labs
priced per deal
One lab gets sole access for a set term
per new batch delivered
License new records as you create them
Typical market ranges paid to the data owner for comparable licenses. Not offers or guarantees. The lab pays you; our commission is agreed with you upfront and comes out of the deal.
For groups, PE portfolios and companies with several distinct datasets. One process, one point of contact, each dataset matched with its own best buyer.
The same factors come up in every data deal, whatever the industry. The more of them your records have, the higher in the range your estimate lands.
Years of continuous records show how decisions played out. Every estimate starts by asking how far back your data goes.
More contributors means more varied, realistic work. Headcount is one of the first inputs buyers look at.
Tickets linked to code, deals linked to emails: linked systems let a lab rebuild a whole workflow, and pay more for it.
Judgment calls, escalations and fixed mistakes are what public data lacks, so they carry a premium.
Data that can't be scraped, bought elsewhere or generated synthetically earns the top of the range.
Records produced by lawyers, clinicians and engineers replace work labs otherwise pay experts by the hour to recreate.
A smaller dataset with clear ownership beats a large one with doubts. Buyers check provenance before closing.
Inputs linked to results (approved, failed, won, lost) can be turned into graded tasks and tests.
Step-by-step records of tool use can become training environments for AI agents, one of the fastest-growing areas of demand.
Text with images, video, CAD or telemetry from real production use is worth more than any single format alone.
Every option, side by side
| Module | Feature | As-is | RecommendedPackaged | Exclusive | Ongoing |
|---|---|---|---|---|---|
| Deal | Best-buyer matching | ||||
| Licenses to several labs | |||||
| Payment | One-off | One-off | One-off, premium | Recurring | |
| Preparation | De-identification | ||||
| Structured schema | |||||
| Tasks & trajectories | Optional | ||||
| Datasheet for disclosures | |||||
| Your control | You approve every buyer | ||||
| Sample review | |||||
| Ownership stays with you | |||||
| Cost to you | Upfront fee | None | Quoted first | Quoted first | Quoted first |
| Osventa commission | From the deal | From the deal | From the deal | From the deal |
No. They're typical market ranges for comparable licenses, to help you judge whether a conversation is worth having. Your estimate depends on your data's volume, history, rights and current demand.
Models are increasingly trained on how real work gets done: real code reviews, real engineering decisions, real case files. That can't be scraped from the public web, so labs license it from the companies that hold it.
Most deals are a one-off license fee paid on acceptance. Some data, like ongoing usage or new records, can be licensed as a recurring stream.
It's agreed with you in writing before we approach any buyer and depends on deal size and scope. It comes out of the lab's payment, so you never pay us out of pocket.
Still unsure? Ask us directly
Two minutes, no obligation, nothing leaves your systems.
Get an estimate