Supplier Document Intelligence

The important question isn't only what the AI extracted. It's whether someone can verify it.

What it is

A procurement-review prototype that reads supplier quotations and specifications, checks them against each other, flags conflicts, and prepares an approved ERP draft — with the source evidence beside every value.

Who it's for

Procurement and supplier-pricing analysts who prepare the data, and procurement managers who approve it.

Problems it solves

• Quotations and specifications arrive in inconsistent formats; one wrong quantity break or outdated revision causes downstream errors.
• A generic AI extraction shows a neat result, but not where each value came from or whether two documents disagree.
• Re-typing data into ERP is slow — yet automating it must not hide who checked and approved the change.

My role

Requirements analysis · product flow · AI-assisted prototyping · evidence design · testing (client project at YBBE TECH, anonymised)

Status & limits A synthetic-document prototype: no measured extraction accuracy, certified pipeline or production ERP connection is claimed, and nothing is posted, ordered or paid automatically.