Making quality visible via data products

AI Coding meets Code Health 🪄 — with Stuart Caborn

Topic 3 of 8

Published Apr 17, 2026 · Clip 13:20–22:34 (9:14)

Stuart explains that loveholidays builds data products for engineering information (code health, incidents, deployments), exposed via dashboards and, once MCP arrived, queryable through MCP servers so anyone—and any AI—can access them. The poster-child use case was incident analysis: an engineer asked the AI for estimated booking loss during an incident, an expert trader corrected the modeling, the question was replayed to the AI, and the result was distilled into a reusable prompt. Stuart then gives practical advice: learn data products, treat data like a quality API with strong documentation governance, use BigQuery and dbt, expose DBT metadata and join patterns to the MCP, and build evals and feedback loops so data usage can be measured and improved.