Method
The sample checks the factory data. The data describes the lot.
A conventional report tells you what an inspector saw in a sample. ROBION connects that sample to the factory's full-lot test stream, then adds narrow machine checks across every unit.
Three layers of evidence. One reviewed result.
Every unit, factory-run
Factory data
The factory's end-of-line results, captured as a tamper-evident stream.
AQL sample, ROBION-run
Independent sample
Tests whether the factory data and the physical lot agree.
Every unit, machine-run
Narrow 100% checks
Labels, serials, firmware and charge behaviour checked across the whole lot.
Worked example
A clean sample can still be weak evidence.
Change the lot, sample and defect rate to see why sample size and rare-defect rate matter.
Chance the sample contains at least one defect
22.1%
At this rate a clean sample is weak evidence the lot is clean. ROBION also verifies the factory's 100% data.
Charge curve analysisPass
- Sample
- 125 units
- Nominal
- 10,000 mAh
- Median
- 9,842 mAh
- Flagged
- 0
Matching data strengthens the result.
ROBION checks serial continuity, distributions, timestamps and outliers. A mismatch is investigated before the lot receives a result.
After review, one of three results.
Pass
Release
Evidence meets the agreed inspection scope.
Conditional
Correct
A defined issue requires action or documentary closure.
Fail
Hold
Material evidence does not meet the agreed acceptance criteria.
Webhook: inspection.resultPass
{
"inspection_id": "RB-260914-042",
"lot": "CN-SZ-4821",
"result": "pass",
"sample": { "tested": 125, "failed": 0 },
"factory_data_verified": true,
"signature": "ed25519:7f9b...2a11"
}