The audit dossier, ready before the auditor shows up

The problem

Before every audit, someone rebuilds the campaign from scratch.

regulatory council and export-certification audits require evidence that today gets rebuilt manually, risking observations for documentation gaps.

• Regulatory council inspections and export certification audits ask for evidence: which grapes went into which tank, who cleaned it, when it was racked, which batch was bottled with which back-labels. Export certifications add their own requirements on top, often for the same batches.
• That evidence exists, but it is scattered across paper logs, spreadsheets, the ERP and people's memory, and nobody owns the whole picture. Each department keeps its own part in its own format.
• So the days before the auditor arrives are spent rebuilding it, and whatever nobody wrote down at the time stays as a gap.
• Those gaps are where observations for incomplete documentation come from, and they put origin certification and export contracts at risk.

How it fits the IRIS system

Edge captures each critical change; Agents files it by certification standard.

Edge (captures photo plus data per change) plus Agents (organizes by certification standard, generates a structured dossier per period/campaign). Zero manual post-hoc work.

Evidence written at the moment of the change cannot have the gaps of evidence rebuilt weeks later. The audit stops being a project and becomes a download of what already exists, filtered for whoever is asking. And because every record carries its photo and its signature, the question of who did what and when is answered by the dossier itself, not by the people who happened to be there.

See the full IRIS architecture →

Before and after

Rebuilt traceability versus the dossier kept live

Aspect	Today	With iLEAN Agents
Preparing a council inspection	Days of rebuilding	None, the dossier exists
Evidence per critical change	A note, if someone wrote it	Photo, log and sign-off
Origin and export standards at once	Separate folders for each	Same evidence, filed per standard
Documentation gaps at the winery	Found by the auditor	Flagged when the change happens
Organized by	Whoever prepares the audit	Period and campaign, automatically
Risk of an observation	Present in every audit	Eliminated for incomplete documentation

days of pre-audit rebuilding → zero; risk of observation for incomplete documentation → eliminated.

Impact estimate

Impact estimate per audit — to be validated with your inspection calendar.

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.

• Estimated payback 6-12 months, depending on how often you are audited. Wineries with several export certifications usually sit at the shorter end.
• Days of pre-audit rebuilding reduced to zero, for the council and for every export certification. That time comes back to the quality team every single year. The people who used to rebuild the campaign go back to their own work.
• The risk of an observation for incomplete documentation is eliminated. Each observation avoided is also a follow-up visit avoided.
• And continuous protection of origin certification and export contracts, which are worth far more than the audit itself. An importer that asks for traceability gets it the same day.

continuous protection of origin certification and export contracts. Estimated payback of 6-12 months depending on audit frequency. Estimate to validate.

And the fair question from the production manager

"Isn't this just a document repository with a new name?" — a repository keeps what someone decides to upload, after the fact, with the same gaps. Here each change generates its own record when it happens, and Agents only classifies it: matching a record to the clause of a standard is an anchored task, where the best models drop below 1.5% error [1], and the winemaker signs every entry.

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