THE AUDIT DOSSIER IS ALREADY READY

The problem

The expensive part of an audit happens weeks before the auditor arrives.

The day the IFS or BRC auditor shows up, someone spends weeks rebuilding traceability of the period's format and allergen changes. The auditor asks for evidence, the plant shows loose papers, and any documentation gap becomes a recorded non-conformity.

• When an IFS or BRC visit is scheduled, somebody spends weeks rebuilding the traceability of the period's format and allergen changes. It is always the person you can least afford to take off the floor.
• The evidence exists, but scattered across paper, a shared folder, the ERP and several people's memories, which is not the same as having it. An auditor does not grade intentions.
• The auditor asks for proof of one specific changeover, the plant produces loose sheets from a binder, and that is where a documentation gap becomes a recorded non-conformity. The change itself was done properly; the file simply cannot prove it.
• The non-conformity is rarely about hygiene: it is about being unable to show what was already done correctly, by people who did their job properly.

How it fits the IRIS system

Edge captures, Agents organizes — the dossier closes as the change happens.

Edge + Agents: every critical change automatically generates pre/post photos of each critical point, a decision log, and timestamped signatures, organized by certifying standard (IFS, BRC, halal/kosher).

Nothing extra is asked of the floor, and that is the whole design. The evidence is a by-product of cases 9 and 10 already running, indexed by standard so that IFS, BRC, a halal or kosher requirement and a customer audit all read the same material through different windows.

See the full IRIS architecture →

Before and after

Preparing for the audit versus a dossier that is always closed

Aspect	Today	With iLEAN Edge and Agents
Work before a visit	Weeks of reconstruction	Hours of review
Serving two standards at once	Two separate builds	One capture, two indexes
Proof of one changeover	Loose sheets, if they turn up	Photos, log and signature together
A record nobody ever filed	Found by the auditor	Flagged the day it should have existed
Halal and kosher requirements	A parallel folder	Another view of the same pack
Where the evidence lives	Paper, folders and memory	One downloadable dossier

Weeks of pre-audit reconstruction → zero. Documentation non-conformity risk → eliminated.

Impact estimate

Impact estimate — to be validated with your audit 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 many certification, distributor and private-label audits you take per year.
• The weeks of pre-audit reconstruction go to zero, and they are weeks of your most expensive quality time, spent on paperwork instead of on the lines.
• Documentation non-conformities — the kind you earn by being unable to show good work that was actually done — stop appearing in the report.
• And the certifications that condition access to the distributor stop being renegotiated every year from a standing start, with the file already closed before the date is set.

Continuous protection of certifications tied to the segment; estimated payback of 6 to 12 months depending on audit frequency. Estimate to be validated.

And the fair question from the production manager

"Will an auditor accept photographs generated by software?" — what is presented is not a model's opinion but dated images of a real state plus the signature of the person who approved it, which is more than a ticked box offers today. Where a value has been read rather than photographed, the task is anchored and the best models stay below 1.5% error [1], with the human signature already on the record. In practice auditors ask the opposite question: why the rest of the plant is not documented this way.

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