Real cases

Industrial AI in PET packaging plants

A PET packaging plant lives off one number: how many kilos of resin come in through the silo and how many thousands of sellable containers go out through the dock. Resin is between sixty and seventy per cent of product cost, and everything lost along the way — preform overweight, changeover purge, start-up queue, scrap from defects, held batches — shows up aggregated in a month-end settlement, with no root cause and no time left to correct it.

See the 12 cases, one by one ↓

On top of that structural pain sit the daily ones. The resin dryer and the high pressure compressor, which govern quality and energy cost, have no network connection. The mold change eats hours of availability and sends kilos to regrind, and "ready to start" is a signature with no evidence behind it. The lab works in a spreadsheet the ERP never sees. Bottlers change the schedule by email and the plan finds out late. And with every point of recycled resin added to the blend, process variability rises and so does the documentary burden of the recycled content declaration.

These pages cover twelve IRIS use cases applied to PET converting plants, grouped into the three blocks of the iLEAN architecture: Connect (capturing what is lost today: shop floor paper, unconnected panels, email, voice, lab sheets, resin receiving), Edge (inline AI vision over preforms, bottles and mold changeovers) and Agents (automatic audit evidence and the coordinated resin balance from silo to pallet). Every case is grounded in the trade, with a measurable before and after and conservative return ranges flagged as estimates to be validated.