A buyer texts about a Tesla in San Francisco. Hard filters run first. Jev scores the evidence for up to 30 candidates, and the buyer gets five options. After they pick, weighted rules prepare a deal plan that is not a binding offer.
A dashboard shows the text thread, the tool calls, the short list, and the decision trace.
It was built at the San Francisco Jev hackathon.
What you get
- TypeScript source
- A short list of five from up to 30 candidates
- A dashboard of the text thread and the decision trace






