Source: www.databricks.com/blog/announcing-ga-genie-ontology
Genie Ontology is an agentic ontology that agents build from everyday team work and people govern. The release adds Genie Search for enterprise knowledge, extends learned context to BI tools such as Power BI and to organizational directories, and brings Unity Catalog Domains and Pages to general availability. Ontology Manager, a visual curation interface, is coming soon in Beta.
Genie Search indexes sources including Google Drive, Jira, Confluence and Workday, with SharePoint in Beta, and enforces each source's access controls before returning results. OntoRank ranks learned and modeled knowledge by factors such as authority, certification and recency. Genie MCP makes the same context available to agents such as ChatGPT, Claude and Microsoft Copilot.
In a benchmark with Turing on a simulated Databricks workspace for a fictional UK energy supplier, Genie passed 91% of runs, against 45% and 58% for two frontier agents. Genie averaged 100 seconds and 432K tokens per run, while the other two averaged 435 and 307 seconds and 889K and 1.05M tokens.
Why it matters if you build agents
Developers can point agents at governed definitions and Pages instead of having each agent rediscover business rules on every run. Search results respect existing source permissions, so teams do not need a separate permission model for enterprise search. Ontology Manager, and Tableau and Sigma support, are not yet available, so plan around those gaps.
Get stories like this every weekday
One short email with the day's agent tooling news.