Databricks Unveils Always-On Genie, an AI Agent That Stops Waiting to Be Asked

Databricks unveiled the second generation of its enterprise AI assistant, Genie, on November 11, 2026, at its Data + AI Live event. The new Always-On Genie is no longer a chatbot that waits for a question: it is an autonomous agent that works in the background, manages long-running tasks, delegates work to sub-agents, and proactively brings decisions to people through a new priority inbox and email. The announcement was made in a keynote discussion between Databricks co-founder and chief executive Ali Ghodsi and OpenAI president Greg Brockman.

What Happened

Databricks announced that Genie is now an always-on, autonomous agent for enterprise work. Employees can give Genie a goal and it takes ownership of the work: running jobs on a schedule or in response to triggers, delegating to sub-agents when needed, following up on missing inputs, and reaching out when a person needs to make a decision.

The release has several components. Genie and Genie Ontology are generally available to all customers. Genie Search, Ontology Manager and the new always-on mode are coming soon to preview and beta programmes. Genie Ontology is the agentic ontology for enterprise AI: it automatically maintains an understanding of how a company operates across its data, people and processes, providing the horizontal context that lets Genie give accurate, trustworthy answers and take actions. Genie Search expands to more than 100 enterprise data sources, including the lakehouse and business apps such as Google Drive, Jira and Confluence, with Workday covered and SharePoint in beta.

Genie is governed by Unity Catalog and Unity Gateway, the governance layers Databricks customers already use. Every action is auditable, administrators control permissions, and Genie acts under each user's own identity and permissions, so it sees only the data that person is allowed to see. Admins can restrict which tools Genie uses, require human approval for actions, and set budgets and spending caps.

In a benchmark built with Turing, Databricks says Genie passed 91 percent of task runs, against 58 percent for the stronger of two anonymised frontier agents — the company's own evaluation, not an independent one. The company also says General Motors has deployed more than 1,000 Genie agents, and thousands of customers already use Genie in production.

What Genie Is

Genie began in June 2024 as a conversational data assistant: users asked a question in plain language, and it wrote and ran SQL on governed company data, then returned tables and charts. The November 2026 release is the next step in that progression — from answering questions, to producing finished work, to doing work autonomously.

The key concept is the ontology. Most enterprise AI fails not because the model is weak but because it lacks context: it does not know what a metric means in this company, who owns a process, or which numbers are trusted. Genie Ontology captures that context automatically and keeps it current.

Why It Matters

The launch captures the biggest shift in enterprise software this year: the move from chatbots that answer questions to agents that do work. Every major platform vendor is making the same move, and the competition is now about trust and governance rather than raw model capability. An agent that can act on its own is only useful if a company can control what it touches, audit what it did, and hold a human accountable for the result. Databricks is betting that its governance layer — Unity Catalog, Unity Gateway, per-user permissions — is the moat that makes autonomous agents safe to deploy.

Caveats for buyers: the always-on mode is still in preview or beta, no price was announced, and the product runs only inside an existing Databricks environment.

What This Means for Founders and Investors

For founders building enterprise AI, Always-On Genie is both a platform to build on and a competitor to route around. The platform opportunity is real: Genie's ontology, connectors and governance create an ecosystem where specialised agents, skills and workflows can be sold to Databricks customers. The competitive threat is equally real: any startup whose product is "ask your data a question" now faces a free, built-in alternative from the platform that already holds the data.

For investors, the release is a data point in the race to own the enterprise agent layer. Databricks is a private company valued among the most valuable in the world, and its ability to convert its data-platform position into an agent platform is central to that valuation. If agents that act autonomously become the default enterprise interface, the companies that control the data, the governance and the context will capture most of the value — and Databricks is positioning itself as one of them.

Angeled.com connects founders with investors.