We go to the holders
We approach operating companies, founders and the trustees of companies that are winding down—sources a marketplace or a contractor network does not reach.
For AI buyersStart with the data
Start with the data
your team needs.
Bring us a defined data brief: code, documents, structured datasets, images, audio, video or specialist material. We find companies whose data fits, check it before you spend time on it, and introduce you with the holder’s approval.
Share a buyer briefWhy work with WitburData that isn’t
Data that isn’t
on the market.
We look for data held inside operating and closing companies: collections that are not listed on data platforms or contributor networks.
Checked before you see it
Before proposing an introduction we establish what exists, who can authorize a license and the known restrictions, so your team evaluates real options.
Terms agreed up front
Our referral fee, and when it becomes payable, are agreed with you in writing before any introduction. Platforms that source data for AI labs can work with us too.
Your briefWhat to
What to
include.
The more specific the brief, the easier it is to rule collections in or out.
- 01
Domain & task
The knowledge area and the behavior or capability the data should support.
- 02
Data & context
The data types—such as code, text, structured data, images, audio or video—and the context or metadata required.
- 03
Format & scale
Preferred formats, approximate volume, languages and period covered.
- 04
Evaluation & timing
Acceptance criteria, any approved sample process, budget and timetable.
- 05
Rights & handling
The intended use, required license scope, restrictions and protection requirements.
What makes data usefulWhat we check
What we check
for you.
File counts say little. These are the questions we look at before proposing a collection.
Real working context
Understand how inputs, actions and outputs relate. A code change with its issue and tests, or a measurement with its method and outcome, can carry more context than an isolated file.
Domain knowledge
Identify the specialist methods, vocabulary and judgment the material captures, and who can explain how the data was created.
Traceable origins
Establish the source, coverage, known exclusions and questions about authority, permissions and permitted use before evaluation progresses.
Published examples of buyer requirements: OpenAI’s 2023 data-partnership brief covers text, images, audio and video. Scale’s program describes interest in operational systems and domain expertise. Each buyer sets its own criteria.
Working together
Who does what.
Source & coordinate
Clarify the requirement, assess potential fit, raise preparation questions and coordinate an introduction with the holder’s approval.
Describe & authorize
Explain the collection, establish authority, identify restrictions and decide what can be shared and licensed.
Evaluate & agree
Define the intended use, evaluate suitability and agree license and handling terms directly with the holder.
Fees
How we’re paid.
Witbur’s referral fee is paid by the buyer. We agree the basis and payment trigger with the buyer before an introduction.
We tell the holder how we are paid. A brief starts a search; we can’t promise a match, and license terms are agreed between you and the holder.
Let’s talkWhat data
What data
do you need?
Describe the domain, intended use and the criteria your team needs to evaluate.