Audit Evidence Pack Ready

The problem

The audit lasts a day. Rebuilding the evidence takes weeks.

On the day of the IFS auditor or the kosher/halal inspector, someone spends weeks rebuilding the period's cleaning traceability; documentation gaps trigger non-conformities.

• A functional ingredients plant answers to several schemes at once: IFS Food, the kosher certifier and the halal certifier, plus customers who audit their own suppliers. Each visit looks at allergen control and changeover cleaning first.
• Before each visit someone spends weeks rebuilding the cleaning traceability of the period: checklists, swab results, changeover records and signatures pulled from different binders and spreadsheets.
• Each scheme wants the same evidence organized differently, so the work is repeated for each auditor and each inspector. A plant with three certifications and a handful of customer audits does it several times a year.
• Gaps found during reconstruction — a missing signature, a changeover without a record — become documentation non-conformities on the day, and they are almost always gaps in paperwork, not in hygiene.

How it fits the IRIS system

Edge and Agents — evidence captured at the changeover, filed by standard.

Edge + Agents: every critical changeover generates before/after photos of each point, a decision log and timestamped sign-offs; Agents organizes it by certification standard, ready to download.

Evidence that is closed when the changeover happens has no gaps to rebuild. Agents takes the photos, decisions and signatures that Edge generates and files them against the requirements of each scheme, so the IFS auditor and the halal inspector each see their own view of the same facts. Nobody rebuilds anything, because nothing was left to rebuild; quality reviews and releases the pack, and the auditor downloads it.

See the full IRIS architecture →

Before and after

Audit preparation today versus a standing evidence pack

Aspect	Today	With iLEAN Agents
Pre-audit reconstruction of cleaning records	Weeks	Zero
Before and after photos per critical point	Not taken	Captured at every changeover
Same evidence for IFS, kosher and halal	Rebuilt per scheme	One record, one view per standard
Missing signature on a changeover	Found by the auditor	Detected when it happens
Customer audit of you as a supplier	Another round of binders	A filtered download
JIDOKA AI decision log	Scattered notes	Timestamped and complete

Weeks of pre-audit reconstruction → zero. Documentation non-conformity 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, counting only the audit preparation time removed.
• Weeks of pre-audit reconstruction go to zero, for every certification scheme and every auditing customer, and an unannounced IFS audit stops being a fire drill.
• Documentation non-conformity risk is removed at the source, because gaps are flagged the day they occur and can still be closed.
• And the weeks quality used to spend on binders go back to the floor, where the next allergen changeover is waiting. Preparing for an audit becomes a review of a pack that already exists.

Payback 6-12 months · removes weeks of audit preparation — estimate to validate

And the fair question from the production manager

"Will an IFS auditor or a halal inspector accept evidence generated by a system?" — they accept records that are complete, timestamped and attributable, which is exactly what this produces: images, decisions and signatures from real people at real times. Classifying each record against a scheme's requirements is an anchored task, where the best models drop below 1.5% error [1], and quality reviews the pack before releasing it to any auditor.

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