How KOHO built a Claude-powered self-serve analytics tool on years of dbt governance
Blog post from dbt
KOHO, a Canadian fintech, built KoDa, a self-service analytics tool that lets employees ask business questions through Claude and receive answers from certified dbt data marts in Redshift, using Amazon Bedrock and existing internal infrastructure. Its creators argue that dependable AI analytics depends less on the language model than on years of data governance, including a rebuilt dbt pipeline containing about 1,400 tested and documented models with PII removed. KoDa is limited to read-only analytics queries against governed marts, uses shared company metric definitions, excludes raw data and transaction-level investigation, and provides the same interface to all employees. The team invested heavily in quality assurance through context libraries and an evaluation system containing roughly 375 real analyst questions, while logging all usage and assigning analysts to review answers regularly. KOHO continues to review results used for board reporting and sees its next challenge as improving organizational data literacy so employees understand when self-service answers are appropriate and when to consult data specialists.
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.