A prototype proves the product can exist. A pilot batch proves the process can repeat it — and confusing the two is how ramps go wrong.
Prototypes are built by the people who designed them, with whatever it takes. That is appropriate: the question is whether the concept works. A pilot batch asks something else entirely — whether ordinary operators, working from the released documentation, with production tooling and the real test fixture, can build the same unit repeatedly.
That difference changes what a pilot is allowed to include. No engineer at the bench solving problems in real time. No undocumented adjustment. No hand-selected components. If the pilot needs any of those, the finding is not "the pilot passed" — it is "the process is not ready", and that finding is worth more than the units.
A good pilot produces four outputs: units with full test records, a list of work-instruction corrections, a confirmed cycle time, and a set of open items with owners. The last one is the honest part. There is always a list.
Then the pilot is reviewed together before ramp. Reviewing it after the ramp has started converts a manageable set of corrections into a running change on units already in the field, which is the most expensive way to learn what a pilot was supposed to teach.
— GANI Contract Manufacturing engineering team