
Databricks launches second-generation Genie with always-on agents in preview
The AMW Read
Background agent execution and governed data access extend a major enterprise platform's agent offering, but the always-on mode remains in preview.
Databricks launches second-generation Genie with always-on agents in preview
Databricks has announced a second generation of Genie, its AI assistant, designed to divide long tasks among parallel sub-agents and seek human decisions through an inbox or email. Genie and Genie Ontology are generally available, while Genie Search, Ontology Manager, and the always-on mode remain in preview or beta. The always-on feature is therefore not yet generally available.
The launch puts Databricks in competition for enterprise agent workflows by using its existing data and permissions layer as the foundation. Genie Ontology maps business metrics, definitions, people, and processes; Genie Search is intended to connect more than 100 data sources. Databricks says agents follow existing Unity Catalog access rules, while Unity Gateway allows administrators to set spending caps and approval rules. The company also offers MCP-based Genie plugins for ChatGPT, Claude, Microsoft Copilot Cowork, and Grok Bot, giving it a route into workflows across several assistants.
For builders and investors, the key question is whether governed access and background execution deliver reliable value beyond a chatbot interface. The preview status limits what customers can currently assess about always-on operation. Pricing is usage-based with no per-seat fee, but the source provides no rates, leaving the cost of sustained agent workloads unclear. Databricks says its search approach reduces tool calls and token use; the article provides no independent measurement of those savings.


