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 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.
