Decision systems · Local experiment
DemandWeave
Match small markets under real constraints.
A bounded exact matcher for synthetic buyer groups and merchant production batches, with budgets, deadlines, tier prices and capacity constraints.
The question
A greedy allocation can consume capacity that a better combination would use to serve more buyers. Unknown preferences also must not become assumed consent.
How it works
- Maximize buyers served before considering catalog savings.
- Use bounded exact search and compare with fixed-order and best-order greedy policies.
- Rank unresolved substitution questions by conditional improvement without accepting the substitution on a buyer’s behalf.
Evidence & results
- Stored comparison covers 200 seeded synthetic eight-buyer markets; exact matching served more buyers on 18 of those paired markets.
- A separate assignment oracle checks 60 seeded five-buyer instances.
Scope & next steps
- Small-instance reference solver, bounded to 12 buyers and six offers.
- No live market, customer savings, LLM capability or real payment outcome is demonstrated.
Go one level deeper.
Ask about the implementation, tradeoffs, or my contribution.
Explore with the portfolio guide