How brokering an AI data deal works

For companies licensing data

1. 1

   Free consultation

   Thirty minutes on what you hold, which systems it sits in and how many years it covers. No data leaves your systems. We tell you honestly if we think buyers will not be interested.

2. 2

   Valuation and rights pre-check

   A written view of likely buyers, a realistic range based on published buyer terms and closed deals, and the rights issues that need clearing first. Our fee is set out in writing at this point.

3. 3

   Data manifest and exclusions

   We help you list each data set by system, record count and date range, and decide what to leave out: HR files, privileged material, client work you cannot license. See how to build a data manifest.

4. 4

   Vetting

   We run the eight checks in our vetting standard. Most issues found here can be fixed by excluding records or getting a client's consent.

5. 5

   Redacted sample to verified buyers

   Only buyers who have passed our checks see a representative, de-identified sample, under a mutual NDA. You approve each buyer before they see anything.

6. 6

   Offers compared on the same terms

   Offers are only comparable when they cover the same records, uses, exclusivity and term. We put them side by side so the highest headline is not mistaken for the best deal.

7. 7

   License signed directly with the buyer

   You contract with the buyer, not with us. Your own lawyer reviews it. We help you negotiate scope, deletion, de-identification certification and payment triggers.

8. 8

   Delivery, acceptance and payment

   A fixed export of only the approved data goes through the buyer's approved transfer method. The buyer checks it against agreed acceptance tests and pays on the agreed schedule.

For AI teams buying data

1. 1

   Free consultation

   What the model needs to learn, what data would teach it, and the license terms you need. We turn that into a written requirement.

2. 2

   Buyer verification

   We verify your company identity, intended use, security and retention commitments and onward transfer limits. You do this once.

3. 3

   Sourcing and vetting

   We search for sellers whose records match, and put each through the same eight-point vetting standard before you hear about them.

4. 4

   Shortlist with vetting files

   For each seller: rights evidence, provenance record, de-identification method, manifest and proposed license terms.

5. 5

   Samples under NDA

   Representative samples, drawn at random and labeled with how they were drawn, so you can test fit with your own pipeline.

6. 6

   Contract, delivery and acceptance

   You contract directly with the seller. We help agree acceptance tests, staged payments and what happens if prohibited material turns up after delivery.

How long does a deal take?

Buyers' own published timelines give a guide. Handshake AI says pricing comes in 12 to 24 hours and its process takes 7 to 14 business days. Polyshares says it makes offers within one to two weeks. micro1 allows up to 30 days to approve delivered data and pays within 30 days of invoice. Allow six to twelve weeks from first consultation to payment for a typical company archive, longer if client consents are needed.

What we never do

• We do not take custody of your raw data. Buyers see redacted samples under a mutual NDA, and full data moves only after you sign a license directly with the buyer.
• We are independent of the buyers we work with. We do not own an AI model and we do not buy data for our own use.
• We never send your company name or any record to a buyer you have not approved.
• We never give a buyer's tools direct access to your live systems.