How KOHO built a Claude-powered self-serve analytics tool on years of dbt governance

KOHO, a Canadian fintech, developed KoDa, an internal self-service analytics tool that enables employees to ask business questions through Claude and receive answers drawn from governed dbt data marts in Redshift. Its creators argue that the key challenge was not AI integration but the years of data governance, organizational trust, standardized metric definitions, and dbt infrastructure that preceded it, including a 2022 rebuild that produced roughly 1,400 tested and documented models without PII. KoDa is limited to read-only analytics, certified marts, and shared company definitions, using existing infrastructure including Amazon Bedrock, Claude, and MCP connections. The team devoted substantial effort to evaluation through context libraries, an “eval council,” and a collection of about 375 real analyst questions, while logging and reviewing live outputs to assess quality and maintain trust. KOHO’s leaders contend that future progress depends less on building AI tools than on strengthening evaluation practices, teaching employees when to trust automated answers, and improving company-wide data literacy.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.