Databricks maps enterprise context into Genie Ontology for AI agents

Context as the enterprise bottleneck

Enterprise agents often fail because business context is scattered, not because the base model cannot write a fluent answer. Databricks used a June 20 tweet to frame Genie Ontology as an automatic context layer that extracts knowledge from tables, queries, dashboards, pipelines, and connected apps, then organizes it into a living graph of how a company works and what its data means.

The substantive line in the tweet is that Genie has context about where to look, what to trust, and how to answer in a way that reflects how the company actually uses its data. Databricks says that includes metric definitions, business terms, unique calculations, and relationships between concepts, metrics, tables, and teams. The linked blog places Genie Ontology alongside Genie One and Genie Agents, making it the grounding layer for agents that reason over structured and unstructured data.

Databricks’ official account typically carries Lakehouse platform updates and Data + AI Summit material. This post matters because it moves the agent conversation from prompt interfaces to governance and trust. The blog says Genie Ontology weighs source authority using an approach similar to PageRank, considering who authored a definition, how often it is used, its ties to certified assets, and freshness. It also says permissions are enforced so users only see content they are allowed to access.

What to watch

The hard test is whether automatically extracted ontology reduces metric conflicts in real companies. If revenue, churn, or active-user definitions vary by team, an agent can still sound confident while answering the wrong question. The source tweet is on X, with the product context in the Databricks Blog.