BOMMatcher

AI matching of bills of materials against supplier catalogues, for construction companies.

Weeks → hours: to match line items
BuilderTrend: integration

ROLE Product, design, engineeringSTACK LLM · Vector search · PythonTIMELINE 2025

I want something like this Process automation

THE PROBLEM

Estimators matched hundreds of line items by hand - it took weeks and it came back with errors.

WHAT WE BUILT

We parse the BOM, match line items against supplier catalogues with an LLM plus vector search, and return a priced estimate with alternatives. Integrated with BuilderTrend.

Live screenshots are on the way

Screens · 01/04

Home screen

What we shipped, in detail

Quick mode handles fast text-plus-image matching for daily line items; thorough mode runs a deeper supplier search across more sources for high-value items, where a few extra minutes saves real money on a large order.

Every match comes back scored, so estimators triage at a glance - high-confidence matches auto-approve, low-confidence ones get flagged for the much smaller human review pass.

Each line item returns the top five suppliers ranked by price, availability and lead time in one view, instead of five browser tabs per item.

Image search finds candidates from a photo or render even when the spec has only a partial part number - a common gap in real-world designer files.

The system generates a full sourcing report in PDF or Excel, formatted to drop straight into the client's existing procurement workflow, replacing ad-hoc spreadsheets.

Previous case

Sarah - AI Voice Banking Assistant

Next case

YM Admin Panel

I want something like this.