No start-up until cleaning is camera-confirmed

The problem

Every variety change rests on a signature with nothing behind it.

every variety or range change is a moment of silent risk — responsibility rests on a human sign-off with no structured visual evidence, right in the middle of peak campaign.

• Switching variety or range in a tank or on a line requires certified cleaning. If wine residue or tartrate stays behind, the next batch starts mixed.
• Today the cleaning is signed off on a checklist. The signature says someone looked; it does not show what they saw.
• Those changes pile up right in the middle of peak campaign, when the cellar is shortest of time and hands and every tank is needed again the same day. Under that pressure, cleaning is exactly the step that gets rushed.
• And when a reserva shows a fault months later, the investigation finds a ticked box and no evidence of what the tank looked like. The person who signed may not even remember that change among the dozens of the campaign.

How it fits the IRIS system

Edge with JIDOKA AI — start-up stays blocked until the image matches and the winemaker signs.

Edge plus JIDOKA AI. Cameras with CNNs trained to recognize "clean vs. wine residue". Start-up blocked until OK plus the winemaker's sign-off.

The camera does not replace the winemaker's judgment; it gives that judgment something to look at. The sign-off stays human, but it is now made on visual evidence at each critical point instead of on trust in whoever cleaned the tank. And every sign-off, with its image, becomes part of the evidence the regulatory council can ask for later, without anyone having to prepare it.

See the full IRIS architecture →

Before and after

Blind sign-off versus camera-backed sign-off

Aspect	Today	With iLEAN Edge
Proof that the tank is clean	A ticked checklist	Image compared with the clean reference
Wine residue on walls or valve	Seen if someone looks closely	Detected before filling
Risk of variety mixing	Silent	Blocked at start-up
Who authorizes the fill	Whoever signs the sheet	The winemaker, on the image
Reserva and gran reserva batches	Exposed at every change	Protected at every change
Evidence for the council	None	Stored per tank and per change

blind sign-off with risk of variety mixing → sign-off based on visual evidence at each critical point.

Impact estimate

Why the payback depends on how often you change variety.

The block below is an estimate to be validated against your plant's actual data. We put it forward so the committee has an order of magnitude; we refine it during the assessment.

• No fixed payback range: we do not set one, because the return depends on how often your tanks and lines change variety or range.
• It protects the quality of the highest-value ranges, reserva and gran reserva, where one mixed batch costs the most. A reserva batch compromised by residue from a young wine loses the value of years of aging.
• It protects origin certification, because each change leaves visual evidence instead of a signature alone.
• The estimate is built with your own number of changes per campaign before anything is committed. Wineries with many varieties sharing few tanks usually see the case pay faster.

protects the quality of the highest-value ranges (reserva/ gran reserva) and origin certification. Payback depends on change frequency. Estimate to validate.

And the fair question from the production manager

"What if the camera blocks a clean tank during harvest, when we cannot afford to wait?" — comparing a known tank against its own clean reference is an anchored task, where the best models drop below 1.5% error [1]. When the image is doubtful, the system does not block in silence: it shows the winemaker the side-by-side comparison, and their sign-off decides.

^[1] OpenAI paper "Why Language Models Hallucinate", 2025 — on the reliability of AI in anchored tasks.