The audit dossier builds itself, changeover by changeover

The problem

The cost of the audit is not the day of the audit.

With double audits — the certifying body plus each private-label client auditing before renewing a contract — the documentation load multiplies compared to a single-brand plant. Rebuilding it manually exposes the plant to nonconformities from documentation gaps.

• The evidence exists, but it is spread across paper cleaning records, spreadsheets, the ERP and the memory of the people who were on the line that day. Nothing is missing, and nothing is together.
• With double audits — the certifying body plus every private-label client auditing before renewing a contract — the documentation load multiplies compared to a single-brand plant.
• Rebuilding the traceability of a period's formula changes can take days of the quality team's time, and it always lands at the worst moment: peak season, both lines flat out and the same team having to run the plant.
• And the reconstruction inherits the gaps: what nobody wrote down at three on Tuesday afternoon does not appear, and that is where documentation nonconformities come from. The auditor does not find a process failure, they find that you cannot prove the process happened — which, contractually, is the same thing.

How it fits the IRIS system

Edge and Agents — every changeover builds its dossier as it happens.

Edge plus Agents: every critical changeover automatically generates before/after photos of the critical points and a timestamped, signed decision log. Agents organizes the dossier both by certifying standard and by private-label client, downloadable from the auditor's web portal.

Every critical changeover leaves before and after photos of the critical points, a decision log, a signature and a timestamp at the instant it occurs. Agents only reorganizes that same evidence the way the certifying body or each private-label client asks for it. Nobody prepares an audit again: they review a dossier that is already built.

See the full IRIS architecture →

Before and after

Today's preparation versus the permanent dossier

Aspect	Today	With iLEAN Agents
Preparation before the audit	Days of reconstruction	Hours of review
For every client that audits	The whole thing repeated	Same evidence, another view
Where the proof of a change is	Paper, spreadsheet, ERP and memory	One downloadable dossier
A documentation gap	Found in front of the auditor	Detected when it occurs
Before and after photo of the change	Does not exist	Automatic, signed and timestamped
Private-label contract renewal	Depends what turns up	With the file already ready

Days of pre-audit reconstruction → zero. Documentation nonconformity risk → eliminated.

Impact estimate

Impact estimate — to be validated with your numbers.

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 the frequency of certifying-body and client audits.
• Days of reconstruction before each visit turned into zero, at peak season as well as any other week.
• Documentation nonconformity risk eliminated, because the gap is detected the day it occurs.
• And continuous protection of the certifications and the private-label contracts that depend on the audit coming out clean.

Continuous protection of certifications and private-label contracts. Estimated payback 6-12 months depending on audit frequency. *Estimate to validate*.

And the fair question from the production manager

“Isn't this just another document manager, and we already have those?” — the difference is direction. A document manager stores what somebody decides to upload after the fact, which is why it inherits the same gaps. Here the file closes at the instant of the formula change, with the evidence Edge and Connect capture, and that capture is an anchored task with the best models below 1.5% error [1]. Agents only reorganizes what is already certified.

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