Plan multiple search queries instead of betting the sourcing result on one wording.
PROJECT / 04 · Procurement / product intelligence
AI Product Sourcing Agent
The AI Product Sourcing Agent is a procurement intelligence layer that compares marketplace offers on consistent evidence instead of treating search results as interchangeable products.
normalizededuperank
CAPABILITIESWhat it does,
What it does,
without the theatre.
01Query planning02Marketplace abstraction03Offer normalisation04Cross-market deduplication05Evidence-based ranking06Supplier/product provenance
Cross-marketplace sourcing creates a comparison problem: listings use different currencies, units, terminology, supplier evidence and duplicated offers, so search volume alone does not produce a procurement decision.
HOW I BUILT AROUND IT01 02 03 04
Normalise offers into a shared product/supplier model.
Deduplicate repeated or near-identical listings before ranking.
Rank with observed evidence and preserve provenance for the operator.
ARCHITECTURE
What the system is meant to preserve.
- 01Evidence before enthusiasm
- 02Comparable units before ranking
- 03Provenance beside every candidate
- 04No fake supplier certainty
Is it tied to one marketplace?+
No. The public engine is marketplace-agnostic and normalises evidence from different sources into a comparable model.
How does it handle duplicate offers?+
It includes cross-market deduplication so repeated or near-identical offers do not dominate the ranked candidate list.
Does ranking equal supplier verification?+
No. It ranks candidates from observed evidence and provenance; supplier verification remains a separate procurement step.