Quotes for test programs, drafted from the customer's RFQ and spec.

Today the RFQ lands with whoever quotes. They read the spec for the methods it calls up, count specimens across every condition, check each method against your scope, look up prices and estimate when results can go out, often starting from an old quote for a similar program.

In the proof of concept, AI does that reading on copies and drafts the quote: each method, its specimen count, accredited or not, the list price and the turnaround. A line your list doesn't price stays blank for the lab manager.

A quote is a reviewed document. ISO/IEC 17025 puts requests, tenders and contracts through one review: the requirements are defined and understood, the lab has the capability and resources, and the methods chosen can meet the customer's needs. For routine customers that review may be simpler. A new customer's first program gets the full read, so each draft line points to the part of the spec it came from.

Accredited or not, line by line. A testing lab's scope lists its accredited methods. A calibration lab's scope goes further: under ILAC P14 each entry gives the quantity measured, the method or instrument type, the range and the uncertainty. A gauge whose range runs past your scope, or a method your scope doesn't list, can still be quoted, just not as accredited work. Accreditation bodies set quote rules too. NVLAP says a proposal using its symbol must state the lab's current status and give a copy of or link to the scope, and its symbol stays off a quote where none of the work is in scope.

The spec leaves numbers to someone else. ASTM B117 says it sets neither the specimen type nor the exposure period; those come from the product spec or an agreement with the customer, and the exposure period drives the turnaround. Specimen counts multiply: the number per condition, times each temperature or orientation the spec names, plus any retests. When an RFQ cites a method your lab considers out of date or wrong for the job, the standard says the customer is told, so the draft flags it rather than quoting it quietly.

Three lines that get left off. Work sent to another lab is named to the customer, who has to approve it. A customer who wants to watch needs witness points agreed, and the standard's note on working with customers names reasonable access for that. A request for fewer specimens than the method asks for is a deviation the lab can accept only if the results stay valid. Differences between quote and order must be settled before work starts, so a vague quote comes back at receiving.

Price is held tightest. The draft uses your list and past quotes and nothing else: per-specimen rates, setup charges, report fees and rush charges only where your list carries them. A discount, a bundle or a test you have never priced stays blank. Your reviewer will check that totals add up, that each count matches the spec, and that nothing outside your scope is marked accredited.

With a yes from your reviewer, new RFQs get drafted the same way for about 30 days.

No quote reaches a customer until the lab manager approves it, and every blank price is filled in by your people, never by the AI.

The side-by-side carries on through the run. A second job waits until this one is judged.