RFQ Sourcing

Given vendor quotes for a requisition, the engine filters out vendors below the SLA bar, scores the eligible ones on a price / lead-time blend, and recommends a winner. This mirrors the Onoots OS matching + recommend, human confirms pattern (like Dreams ↔ inventory in Proptech): the engine ranks; a human raises the PO.

Scoring

Each eligible quote gets a blended score where lower is better:

score = priceWeight × norm(price) + (1 − priceWeight) × norm(leadDays)

• priceWeight (default 0.7) weights price vs lead time; 1 = price only.
• norm(...) min-max normalizes each dimension across the eligible quotes.
• Vendors with rating < minRating are ineligible (score NaN, sorted last, annotated “Below min rating”).

The result returns the full ranked list, the winner (best eligible), and savingsVsHighest — money saved versus the most expensive eligible quote.

Worked example

Demo RFQ for requisition REQ-7741 (raw stock, 40 units):

With minRating = 4, Apex Wholesale (3.2) is excluded despite the lowest price — the SLA bar wins. Among the eligible three, the priceWeight = 0.7 blend balances Brightline’s lower price against Cedar’s faster lead time and Northgate’s mid profile.