The buyer I never approached runs a factory.
The last experiment ended on a gap: a buyer for whom a widely held answer is worth as much as a private one was never tested. Industrial data that has to convince someone outside the plant may be that buyer. This is the hypothesis, written down before any of it is built.
Where the last experiment left a gap
When I recorded that agents on Base were not coming back to buy specialized checks, one argument refused to resolve. A shared, cached answer is worthless to a buyer whose margin comes from knowing what others do not. But it says nothing about a buyer whose problem is the opposite: being believed. A compliance function, an auditor, a counterparty. For them, an answer everyone can verify is worth exactly as much as a private one. Possibly more.
None of those buyers were ever approached. That was a gap in the experiment, not a finding of it. This note is about closing it in a place I know from the other side of my work.
Why a factory
Outside Tereno, I work as a consultant on AI and optimization for industrial operations, under the name .regeen: process optimization, OEE, predictive models taken to production. Inside a plant, data is trusted by default. The historian says what the line produced, and the people who run the line act on it.
The moment that data has to leave the plant, the trust does not travel with it. An equipment maker billing per hour of productive use and a customer who disagrees about how many hours there were. A foreign buyer that needs proof of origin and of how a product was made. A certifier, an insurer, a lender tying a rate to an efficiency claim. Today that boundary is crossed with PDFs, spreadsheets and site audits.
Industrial data becomes worth paying for at the moment it has to convince someone outside the plant.
Where it could show up
- 01Pay-per-use equipment. The maker sells productive hours or good parts instead of the machine. A production record both sides can verify becomes the basis of the bill, and a payment rail built for machines can settle per unit without anyone issuing an invoice.
- 02Origin and environmental claims. Pulp, paper, food and chemicals increasingly have to prove to outside parties what was produced, with what inputs, and when. A verifiable record of what the plant measured, and that it was not changed afterwards, is the same shape of product as a Tereno evidence artifact.
- 03A cooperative OEE benchmark. Every plant wants to know whether its OEE is good for its sector, and none wants to open its data. Contribute anonymized figures and earn credit; query the benchmark and pay per call; the credit flows to whoever produced the data. That is the reuse mechanism Tereno already runs, with a different kind of fact.
- 04Plant agents buying outside evidence. An optimizer that pays for an input price, a weather window or a supplier's lot quality before it recommends a change. This is the original Tereno thesis with a buyer that has a budget line.
Why it might be wrong
- 01Operational networks in plants are isolated on purpose. Anything that needs the control layer to talk to the internet will not be installed.
- 02Industrial legal and IT teams are allergic to anything that sounds like crypto. If the chain is visible in the pitch, the conversation ends.
- 03Two parties that already share a contract do not need a chain. A signed database solves their problem. A shared ledger only earns its place when several parties need to agree and none of them should be the referee.
- 04Settling in USDC is still awkward to book in Brazil, where most of my industrial contacts are.
- 05Industrial sales cycles are measured in quarters, not weeks. This is slow to prove either way.
The smallest test
One capability, provisionally industrial.attestation. It receives a production or OEE report, keeps nothing but its fingerprint, and returns a timestamped receipt. Later, anyone holding the same report can check that it is the one that was attested, and that it was not altered. No raw data leaves the plant, nothing touches the control system, and the chain stays underneath, where nobody in the meeting has to see it.
| Question | Confirms | Kills |
|---|---|---|
| Does anyone outside the plant care? | A third party verifies a receipt it did not create, without being asked to by me. | The conversations listed before starting end without one party wanting to verify. |
| Will they pay for it? | A verifier or an issuer pays for a second attestation after the first one. | Interest exists only while it is free. |
| Does the chain matter? | More than two parties rely on the same receipt. | Every use case collapses to two parties and a signed file. |
Two sides of the same work
Tereno is where verification and reuse live. .regeen is where I meet the plants that might need it. The industrial side gets no chain in its vocabulary: it gets verifiable records, receipts and an audit trail. If the hypothesis survives, both describe the same product from opposite ends.
This is a hypothesis, not a result. Nothing described here has been built, sold or tested with a customer yet. When it has, the outcome will be published here with its denominator, whichever way it goes.