The Databricks Genie AI agent is gaining autonomous capabilities that allow it to manage complex business tasks over extended periods. Announced Wednesday, Nov. 11, the upgrade is designed to help companies identify operational problems and act on information across their internal systems.
Databricks CEO Ali Ghodsi already uses the technology to analyze employee feedback, review unsuccessful sales pitches, and prepare for meetings. The agent can surface problems that might otherwise remain buried in company reports.
Databricks Genie AI Agent Handles Multistep Tasks
The second-generation Genie can complete multistep tasks without constant human direction. Users can assign objectives, and the agent monitors progress, delegates work and requests decisions when necessary.
Ghodsi demonstrated how Genie examined employee satisfaction survey results. It identified areas with poor scores and evaluated management proposals intended to address those concerns.
The agent also reviewed an unsuccessful sales presentation. It identified customer requirements that the sales team had not satisfied, providing information that could guide future negotiations.
Ghodsi uses Genie to prepare for interviews and meetings by reviewing previous conversations and relevant business information.
He believes these capabilities become particularly useful as organizations grow. Databricks has approximately 12,000 employees, making it difficult for executives to maintain direct relationships across the company.
Genie Ontology Provides Business Context
The upgrade relies on Genie Ontology, a system that connects company data with business definitions, processes and organizational relationships.
It helps the agent understand how different departments interpret information. For example, revenue calculations, approval procedures and performance metrics can vary across business units.
Genie Search also connects information from Databricks and external applications, including Google Drive, Jira and Confluence. The system supports more than 100 enterprise data sources.
Databricks says this contextual approach helps agents produce more accurate answers and complete tasks using existing company knowledge.
Unity Gateway Adds Security and Cost Controls
Autonomous agents can generate substantial computing expenses when continuously processing business data.
Databricks addresses this with Unity Gateway, which lets administrators set spending limits, control tool access, and monitor agent activity.
The platform also supports different AI models, including less expensive open-weight alternatives. Companies can select models based on task requirements rather than relying on one system for every operation.
Genie follows existing data permissions through Unity Catalog. Administrators can also require approval before agents perform sensitive actions.
Availability and Enterprise Deployment
Databricks says Genie and Genie Ontology are generally available to customers. However, the new always-on autonomous capabilities, Genie Search and Ontology Manager, are expected to enter preview and beta programs.
The company uses consumption-based pricing rather than a fixed per-user subscription. Actual costs depend on the computing resources and services used.
The update moves Genie beyond answering business questions toward managing ongoing assignments. Its effectiveness will depend on the quality of company data, access permissions and the reliability of autonomous decisions.
