Insights & Updates
AI agent architecture, data governance, and why deterministic data access matters.
We Ran the Same Benchmark as dbt Labs. Here's What They Didn't Measure.
20 analytics questions. 60 agent calls. Two strategies: SQL generation vs. governed metrics. The results tell a story about fabrication, not accuracy.
Read moreAuthor in dbt. Serve through OnlyMetrix.
dbt gets agents to your metrics. OnlyMetrix makes those metrics better. Compile-time safety, autoresearch, and an analysis layer on top of your existing dbt project.
Read moreYour Metrics Haven't Changed. Why Is Your AI Re-deriving Them?
AOV, ROAS, churn rate, LTV. These calculations haven't changed in decades. So why is every AI agent re-deriving them from scratch on every query?
Read moreHarness Engineering for Data Access
Context engineering optimizes what an AI sees. Harness engineering controls what it can do. For data access, the strongest harness removes SQL generation entirely.
Read moreWhy AI Agents Shouldn't Write SQL
AI agents are querying production databases across every company. But letting LLMs generate SQL is a reliability, security, and governance problem that context injection can't solve.
Read moreDeterministic vs. Probabilistic Data Access
Two fundamentally different approaches to giving AI agents access to your data warehouse. One generates SQL on the fly. The other removes SQL generation entirely.
Read more