We built Procunit because visual inspection on fast lines is still mostly done by tired eyes — and recalibrated rule sets that break every time the line changes.

Built by someone who ran the line.

Amanda Okonkwo spent four years managing quality programs at a contract electronics manufacturer in the San Gabriel Valley before co-founding Procunit in 2023. At her previous job, a customer pushed for a 40% line speed increase on a PCB assembly line. Escape rates doubled within a month.

The shop's AOI system was recalibrated six times in a single quarter and still missed connector orientation errors — a defect class that required only 30 seconds to catch visually when the line ran slow enough. The problem was not calibration effort. It was that rule-based AOI encodes static conditions: fixed thresholds, fixed ROIs, fixed lighting assumptions. Push the line faster, change a part fixture, or add a new defect class and the rules break.

Procunit's first model ran on a $600 industrial PC and caught 94% of defects on a 180 ppm PCB line. That number is now above 99% on fully trained production models. We are not trying to replace every quality program — we are solving one specific problem: inspection that keeps up with the line and learns from your team's accept/reject history instead of requiring a vision engineer every time conditions shift.