AI-Assisted Coach Matching

A corporate coaching provider matched clients to coaches entirely by hand — reviewing enquiries, interpreting needs, and searching profiles across language, seniority, location, and topic. The process was slow, inconsistent, and constrained by strict data privacy requirements.

We built a privacy-aware matching engine combining semantic search, RAG-based retrieval, keyword scoring, and proximity signals. The system surfaces ranked coach shortlists from unstructured, multilingual client inputs — with every match explainable and reviewable by a human operator.

Matching time dropped from 7 days to seconds. Match quality became measurable for the first time via rematch rate, acceptance rate, and override rate. The operations team shifted from manual searching to expert review.