building which is a talent sourcing toolkit that runs inside Claude, ChatGPT, or any other AI assistant.

building https://sengi.ai/ which is a talent sourcing toolkit that runs inside Claude, ChatGPT, or any other AI assistant. It searches LinkedIn and the wider web, turns messy public profiles into structured evidence, screens candidates against a hiring brief, and gives recruiters a ranked longlist with Strong, Partial, and No fit categories. I want to test Jev for typed per-requirement judgments, relevance scoring, and confidence-based routing. I’d compare it with OSS models I have under the hood there now on identical synthetic or approved data and I already have evals for measuring recall, precision/F1, calibration, latency, cost, and evidence quality, so super-excited to get an access!