Decide what to do with your company’s records.
AI labs, and the firms that supply them, now pay companies for their operating records: support tickets, approvals, project histories, written procedures. We show you whether yours qualify, which route fits your plans for the company, and what to check in an offer before anyone sees your records.
Your answers. The link keeps them.
Step 5 of 5
Talk to us about your records
Send your answers with the company’s name and site. We reply with the route we’d take and who we’d introduce you to.
Your result · updates as you answer
Worth a look
Some buyers would look. The gaps below decide what they offer.
53 / 100
How the score is built
- Years of records × 0.15
- 100
- People doing the work × 0.10
- 85
- Records of judgment × 0.25
- 10
- How each job ended × 0.15
- 60
- Written procedures × 0.10
- 35
- Ownership and sensitive data × 0.15
- 30
- Buyer interest in your industry × 0.10
- 100
What would raise it
- Keep the record of decisions: approvals, edits, tickets, and why each was resolved.
- Confirm who owns the records and what customer contracts allow, and set aside personal and regulated data.
- Write down how the repeat work gets done.
A readiness score, out of 100. It is not a price: buyers price records only after they read them. Each part counts for its weight, shown beside it. The weights are ours, published so you can check the arithmetic.
What buyers publish
Checked Nov 5, 2026
Published per-company figures
$10K to $4M+
The span of what 7 programs publish for a company like yours.
These are the programs' own advertised figures. None of them is an offer, and no buyer has seen your records.
Where these come from
- Handshake AI data licensing $100K to $4M+ payouts
- micro1 Company Data Partnerships $200k-2M+ headline; $100K-500K, $500K-2M and $2M+ by tier; payments typically $150K to over $2M
- Polyshares data licensing Typically $100K to $2M, paid by us to the company
- Replay data licensing $10K - $100K for 20-50 employees; $100K - $1M for 50-250; $1,000,000+ for 250+
- Scale AI Data Partnerships $10K-$1M+ illustrative value per data partnership, scaling with cadence
- SourceX data licensing Estimated licensing payout based on company size (for example $30K–$180K for 50–99 employees); exact payout fixed in a signed Dataset Deal Confirmation
- Turing, Project Lazarus up to $1M
Every program, with its terms and its source
Fifteen to twenty minutes on what these figures mean for your records, your rights and the rest of your plans.
Best routes for your plans
1License a defined slice65
One buyer licenses records you choose, from a set period, for a set term or use. You keep the records and the right to license them again.
- +5Your records may qualify
- +10You keep producing records, so licensing a past slice leaves the future yours
If you sell the company: A buyer of the company reads the license and counts one-off license income as non-recurring. A bounded license costs little; a perpetual exclusive one can cost more than it paid.
2Keep them and sell them with the company60
The records stay inside the company and count toward what a buyer pays for it. Nothing is licensed now.
- +10Your plans are open, and keeping the records keeps every route open
If you sell the company: A buyer prices the records with the company, and diligence finds no license to discount.
3License on a schedule60
The history first, then new records each year or quarter, paid as they arrive. It works while the business keeps producing the same work.
- +15You keep producing the same kind of records
- −5Buyers on a schedule want records that already qualify
If you sell the company: Contracted data revenue counts toward the price, if the contract moves with the company.
All 7 routes
4Sell the company to a buyer that prices the records in60
AI-enabled roll-ups and strategic buyers now pay for companies whose records and workflows they can put to work. The records are part of what they buy.
- +10The records keep coming after the sale
If you sell the company: One sale, priced with the records, instead of two.
We also match you with buyers for the company.
5Train a model of your own50
Your team fine-tunes an open-weight model on your own records, in systems you control. The trained weights belong to the company, and the records never go to a lab.
- −10You don’t use AI on your records yet
- +10New records keep improving it
If you sell the company: The model, and the records that keep improving it, belong to the company a buyer acquires.
6Sell the records, or license them forever and exclusively50
One payment. The records’ value to anyone else, including a future buyer of the company, ends with it.
If you sell the company: A buyer counts the payment as one-off income, and diligence finds the contract. Before a sale, it removes something the buyer would have paid for.
7Build a data product40
A valuation and build partner measures what the records support, builds a product buyers ask for, and sells it on license or as a service. You pay fees up front; revenue follows if it sells.
- −20You don’t want to build anything
- +10New records keep a product current
If you sell the company: A product with paying customers is a second line of revenue a buyer pays for.
Each route starts at 50 and moves with your answers. Routes at 60 or more are worth a look.
Your records and your company
Your records and your company are two things a buyer can pay for, and selling the first changes the price of the second. A lab that licenses your records gets better at your work. A buyer of your company reads that license in diligence. So we start from your plans for the company, and price the records inside them.
One payment for your history is the simplest route. If the business keeps producing the same work, that history is the first sale, not the only one: new records can be licensed each year, or built into a product the company owns and sells.
Read the rest
The asset that lasts is the business that keeps producing records, not one export of them. Keep the systems, the people, and the habit of recording each decision and how it ended, and you keep something to license, to train on, or to sell with the company.
A model of your own is a real option now. Open-weight models publish their weights under licenses that let you run and adapt them. Your team, or a partner working in your systems, fine-tunes one on your records, often with small adapter layers rather than retraining the whole model, and tests it on past work it hasn’t seen. Your outcome records (approved, redone, paid) can grade its answers. The records stay with you, the trained weights belong to the company, and the base model’s license sets the terms of use. It won’t match a frontier model at everything; it can do your work your way, and it improves as you keep producing records.
Money from records can be reinvested or taken out. Reinvested in automating your own work, it can raise earnings, and with them what the company sells for. Some buyers, AI-enabled roll-ups among them, now pay for companies whose records and workflows they can put to work, and price the records once, with the company.
Every company in your industry faces the same choice. While few sell, records stay scarce and models stay weak at your work. As more sell, prices fall and the work gets automated for everyone. A bounded license, or keeping the records for a sale, keeps more of your routes open than one perpetual license.
Who we introduce you to
When a route fits, we introduce you to a firm that values company data or a buyer that licenses it, from the data market on our markets page. Before any introduction, we tell you in writing whether that firm pays us, and you decide whether to go ahead.
See the data market