Natural-language portfolio matching · MATCHBeta

Akıllı İlan Arama

Match the buyer: let the client describe what they want, and the right listing surfaces from the portfolio — with why it fits.

What it does

Three-vector hybrid RAG + RRF fusion with a 6-intent explainable score. The client speaks in natural language; the system surfaces the best-fitting listings from the network portfolio with a why-it-fits explanation. Cross-office opportunity, title-deed/risk filter. Multi-tenant by design — proven in a pilot, scaling for franchise networks.

The problem

Matching what a client wants against hundreds of portfolio listings by hand is slow and incomplete.

Highlights

  • Natural-language search — the client speaks, the system matches
  • Explainable score: why this listing fits
  • Cross-office opportunity (network portfolio)
  • Title-deed/risk filter

Stack

3-vektör hibrit RAGRRF füzyonPostgreSQL + pgvectorClaude Sonnet
All products