Data Truffle is a seller-side agent for company data. We price operational records before they go to market, run a competitive process with vetted buyers, and negotiate for the company that owns the records. We are paid by that company, only when it is paid, and we never touch the data.
A new market appeared in the last two years. AI labs, having read the public internet, started paying established companies for the records that show how a business actually runs: tickets, calls, approvals, deals, close checklists. Published ranges for a single company's records run from the tens of thousands to the low millions.
Almost every service that sprang up to meet that market is paid by the buyer. Introduction services pass an owner into one lab's referral program and collect a fee from the lab when it closes. That is not a scandal. It is a side of the table, and it is not the owner's.
The result is predictable. The owner gets one offer, from one buyer, with a short deadline, and no way to know whether it is a third of what a different lab would have paid for the same records. The buyer has done this fifty times. The owner is doing it once.
We built Data Truffle to be the other side of that table: price the records against live demand before any buyer names a number, put several fitting buyers in front of the same manifest at the same time, negotiate the terms that matter alongside price, and get paid only when the owner does. Fast, because the pricing work is done in advance. Clean, because the data never passes through us.
Shawn Livermore builds software products and platforms and is the architect of several SaaS codebases. Data Truffle is the one he built after watching the AI data market take shape with nobody representing the companies whose records were being bought.
There is no invented team on this page. When you book a call, you are talking to the founder. As the company grows, the people on this page will be real ones, with real names.
Every buyer we introduce has agreed to these in writing. A buyer who will not is not a buyer we will represent you to, however large the number.
Verified by us, disclosed to you before any sample moves. Nobody signs with a stranger.
A named method and a named verifier, applied before training. Not a policy page, a clause.
The supplied copy is destroyed at the end of the term, usually 90 days, with a signed certificate.
Your records stop with the buyer. Field-of-use limits and an audit right back it up.
Willing to be named, with permission, to a future seller asking whether they behaved.
No exploding offers. We do not pass along an offer with fewer than ten business days to decide, from anyone.
We do not buy data, train models, or hold inventory. We have no reason to want your records cheap and every reason to want them priced right, because our fee is a share of what you are paid.
There is no listing, no catalogue, no browse button. A marketplace is a venue that charges you to stand in it. An agent works for you, on one deal at a time, and is paid when it closes.
We never receive, store, or transmit your datasets. The manifest is metadata. Samples and licensed records move from you to the buyer directly. We could not misuse your data because we do not have it.
You will not find testimonials, client logos, or deal counts on this site, and we will not invent them. The market is young and we would rather earn that section than fake it. What you can check today:
Seven questions. No dataset. No fee unless a license closes.
Measured from a complete data manifest. A first offer is a process target, not a guarantee; buyers decide eligibility, price, and terms.