You pay for what you keep.

Whatever discovery defines it as, in writing: one finished piece of work and the test it must pass. You agree it before the pilot, and you keep the definition whether or not you go further.

Some of it by software, some by people, and the mix changes as we learn your work. You are billed the same way either way: per unit you accept, with failures reworked at our cost.

Messy inputs are expected. What matters is whether a finished unit can be checked against a source or a rule. Discovery looks at your actual work and tells you in writing whether it fits; if it does not, we say so.

Work with no clear unit, work that cannot be checked against a source or a rule, and work that does not repeat. If nobody can score the output, there is nothing to accept or reject, and we say so at discovery.

A per-unit price needs a real unit. Discovery produces one, in writing, and you keep it whether or not you go further.

One line per type of unit: the number you accepted in that batch, times the unit price. Nothing appears on it that you have not scored and accepted.

No minimum term. The pilot is a fixed batch, the run is ongoing, and you can stop after any stage. The bill follows accepted units at the same price, so a quiet month costs less. Whether a workflow has enough volume to run as units is a discovery question.

You do. You draw the sample and score it against the standard in the contract. A unit is not invoiced until you have accepted it.

A disputed unit stays off the invoice while it is settled. If the standard in the contract does not settle it, the wording was not tight enough, and we fix the wording with you rather than argue the unit.

There is no partly accepted unit. It either meets the standard you scored it against, or it is reworked at our cost, goes through the same test again, and reaches the invoice only once you accept it.

The test does not change after the pilot. Every batch in the run is sampled by you and scored against the same standard, so drift shows up in your scores rather than in a complaint, and the failed units are reworked at our cost.

The standard lives in the contract, so a change is agreed in writing with a date from which the new test applies. Units accepted under the old standard stay accepted.

We do not quote a turnaround before we have seen the work, for the same reason we do not quote a price. Discovery shows how the work arrives and how fast you need it back, and the turnaround goes into the contract beside the standard.

For discovery, the work as it runs today and a sample of real inputs if you have one. For the pilot, a fixed batch of real data, access to the source or rule the work is checked against, and someone on your side with time to score the sample.

No. The pilot runs alongside your current operation on a fixed batch, so you compare results with no risk to the day's work. If you go on to the run, volume moves across at a pace agreed in discovery.

You hold the unit definition and the standard in writing, which is what your own team or another supplier needs to carry on, and you owe nothing for units you have not accepted.

Named people only, with role-based access. Deleted when you ask, and we send you a written certificate. We do not train on your data and we do not share it.

Only the people we name to you in writing for your work, each with access limited to what their role needs. We keep that list current, and nobody outside it has access.

The least that lets us produce the unit and check it against its source. Discovery sets out which systems and what level. The accounts are issued by you and you can revoke them at any time. We do not ask you to change your systems.

Yes to both, before you send us anything if you prefer. We answer the questionnaire in full, in writing, and walk your security team through how your work is handled.

No one we can name yet. We are new. The pilot stands in for a reference: a fixed batch, scored by you, and you can stop there.

You do not have to take anything on trust. Nothing is invoiced until you have accepted it on a sample you drew, failed units cost us rather than you, and you can stop after any stage, including after the pilot.

On the units that pass. A failed unit costs us the rework and earns nothing, so we are paid only for work that clears your standard on your sample.