Source record

Published by LangChain Blog (T1.5 source). The original is at https://www.langchain.com/blog/how-numan-built-an-eval-first-ai-health-coach-with-langgraph-and-langsmith.

Pipeline notes

The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.

Numan built Nu, an AI health coach for weight, hormone, and sexual health management, using LangGraph and LangSmith. The team adopted an eval-first approach with per-category evaluations, simulated conversations, and safety gates to deliver coaching while avoiding regulated medical decisions. Single-turn fixture-based evals proved inadequate for conversational coaching.

Detailed walk-through of eval-first design for a regulated conversational agent, including why static fixture evals fail in coaching and how per-category criteria replaced them.

Cited by

No citations on record.