I Built an AI Complaints Triage App with Snowflake Cortex Code in a Single Session
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
A Snowflake Solutions Engineer built a working AI-assisted complaint triage prototype for a state regulatory agency in a single session using Snowflake Cortex Code (CoCo). The app uses a Cortex Agent with four tools, three RAG indexes (Cortex Search over statutes, case precedents, and document checklists), and Cortex Analyst for plain-English analytics. The frontend is React + Vite + Tailwind with an Express.js backend, deployed on Snowpark Container Services with Snowflake's built-in OAuth for auth. The post covers synthetic data generation, the three-stage human-in-the-loop workflow (intake, investigator review, legal review), and the argument that AI coding assistants make custom internal tools economically viable for small user bases that previously couldn't justify the build cost.
Table of contents
Why Snowflake for EverythingThe Tool That Made This PossibleMaking the Data RealStanding Up the AI LayerGet Michael van Meurer’s stories in your inboxThe DashboardThe WorkflowDeploying It: SPCS and AuthWhat I’d Tell Another BuilderWhere This Kind of App BelongsSort: