AI System for Comparing Supplier Quality Documents with Purchase Orders

The problem

When certificates, purchase orders, batches and technical specifications are held in separate sources, their fields must be matched by hand before a list of document discrepancies can be prepared.

The incoming quality specialist must compare every suggestion in the document discrepancy list with its source record; missing or conflicting evidence must not replace the final decision.

If an approved deviation permit or equivalent document code is not set aside for review, nonconforming material could be accepted or a conforming batch could be held; the record must not proceed without a human decision.

What we set out to improve

• Time taken for a draft document discrepancy list and its supporting evidence to reach the incoming quality specialist for review
• How records set aside for human review because an approved deviation permit or equivalent document code exists are resolved
• Number of suggestions corrected after identifying a risk that nonconforming material could be accepted or a conforming batch held

How the system works

1. 01 Retrieve the certificate, purchase order, batch and technical specification from permitted sources using the record identifier
2. 02 Link source, date and version information to the shared work record
3. 03 Validate fields and units
4. 04 Prepare the document discrepancy list with its supporting evidence
5. 05 Check for an exception: an approved deviation permit or equivalent document code exists
6. 06 Have the incoming quality specialist accept, revise or reject the suggestion
7. 07 Record the approved decision and its source in the processing log

Methods we used

• A source-identified common data schema for the certificate, purchase order, batch and technical specification
• A decision structure that separates deterministic rules from AI suggestions
• A review screen showing the document discrepancy list alongside the underlying source evidence
• A mandatory human review queue when an approved deviation permit or equivalent document code exists
• An event identifier that prevents duplicate processing

Where people stay involved

Data and security

Only the fields required for this decision should be processed from the certificate, purchase order, batch and technical specification. Access to the source systems must be authorised separately from the incoming quality specialist role, and personal data, commercial information and technical secrets must not be passed into the model context unnecessarily. Reading, suggestions and human decisions must be logged separately.

Who this suits

Frequently asked questions