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