Databricks launches always-on Genie, an autonomous AI agent for enterprise work

Databricks on November 11 announced the second generation of Genie, turning its AI assistant from a tool that answers when asked into an always-on agent that carries enterprise work forward on its own.

What changes

Employees give Genie a goal and clear boundaries. The agent then takes ownership of the work, delegates to subagents to run tasks in parallel, tracks progress, follows up on missing inputs and reaches out when a person needs to make a decision. A new priority inbox surfaces what needs attention, and Genie sends email alerts when users are away.

Databricks' examples include marketers asking Genie to monitor content performance, customer success teams having it flag shifts in usage, and executives asking for a weekly brief on revenue trends and top feature requests. Every action stays within existing Unity Catalog permissions and Unity Gateway approval rules and budgets, and is fully auditable, the company said.

Genie Ontology goes GA

Genie's actions are grounded in Genie Ontology, a knowledge layer that learns and maintains a company's definitions, metrics, people and processes. It is now generally available, with new sources including external BI tools such as Power BI and organizational directories. Databricks also introduced Genie Search, which spans more than 100 enterprise sources such as Google Drive, Jira and Confluence.

In a benchmark built with Turing, Databricks says Genie with Genie Ontology passed 91% of enterprise task runs, 1.6 times the rate of the strongest frontier agent, while using less than half the tokens. The figures are the company's own.

Availability

Genie and Genie Ontology are generally available now. The always-on mode, Genie Search and Ontology Manager are headed to preview and beta programs. TechTarget reports that several of the new features were built in response to customer requests.

Sources: Databricks, Databricks blog, TechTarget