Kaelio

Kaelio makes ktx, an open-source context layer built for AI data agents. The platform ingests a company's data stack, including warehouses, BI tools, modeling code, query history, and documentation, and compiles all of it into structured context that agents can search and query at runtime. Instead of giving agents direct table access, ktx provides approved metric definitions, a mapped join graph, and a business wiki stored as plain YAML and Markdown files. Every update is a git-reviewable diff, so data teams stay in control of the logic agents apply. It connects with Snowflake, BigQuery, Redshift, dbt, Looker, and others, and works with agents like Claude Code, Cursor, and LangChain.

For BPO and operations teams that rely on AI agents to surface performance data, ktx addresses a real consistency problem. Agents querying workforce metrics, quality scores, or capacity figures often return different answers depending on which table they hit or which join path they follow. With ktx providing a governed context layer, those agents draw from approved definitions and validated query logic rather than improvising SQL from scratch. Analytics teams can run AI agents for reporting and forecasting without worrying that the underlying metric logic shifts between runs. Because the context files live in git, operations leaders can audit exactly what logic the agent applied, which matters when outputs feed into client reporting or staffing decisions.