Databricks Launches Always-On Genie Agent for Enterprise Work
On 11 November 2026, Databricks announced the second generation of Genie, its enterprise assistant, now positioned as an always-on agent that keeps working on tasks in the background. According to the company, Genie and Genie Ontology are generally available, while Genie Search, Ontology Manager and the always-on mode are in preview or beta. The release adds more than 100 connectors, including Google Drive, Jira and Confluence. Databricks says every agent action is auditable and that administrators control permissions. The agent runs inside a customer's existing Databricks environment, so the product only fits businesses that already use Databricks or plan to adopt it.
In short
- Databricks announced the second-generation Genie on 11 November 2026 as an always-on enterprise agent.
- Genie and Genie Ontology are generally available; Genie Search, Ontology Manager and the always-on mode are in preview or beta.
- The release adds more than 100 connectors, including Google Drive, Jira and Confluence.
- Databricks says actions are auditable and administrators control permissions, but the product runs only inside a Databricks environment.
- Check the current status and pricing with Databricks before planning any rollout.
What happened
Databricks, the data and AI company based in San Francisco, announced the second generation of Genie on 11 November 2026. According to the company's press release, Genie is now an always-on, autonomous agent for enterprise work. It runs in the background on its own compute, and users give it a goal and set its boundaries rather than asking one question at a time.
Trade coverage describes how the agent handles longer jobs. According to iTBrief, the product is aimed at larger firms that want AI to carry longer-running work without exposing sensitive data. SD Times reports that the agent can split work between sub-agents running in parallel, and that each new alert it raises arrives as its own unread email.
Availability: what is ready and what is not
The announcement separates the product into parts that are ready now and parts still in testing. The table below follows Databricks' own wording.
| Component | Status announced on 11 November 2026 |
|---|---|
| Genie | Generally available |
| Genie Ontology | Generally available |
| Genie Search | Preview or beta |
| Ontology Manager | Preview or beta |
| Always-on Genie mode | Preview or beta |
There is a conflict to note. Databricks' own answer pages on Genie Ontology, written earlier in 2026, still describe it as in Public Preview. The 11 November release is the newer statement and calls it generally available. Anyone planning a rollout should confirm the current status with Databricks before relying on it.
Why it matters for businesses that automate work
Most business software still waits for a person to ask. An always-on agent is built to watch for new data, follow up on open items and bring decisions to people. That moves the question from how fast a person can search for an answer to what the agent is allowed to do unattended.
Databricks positions the release around that governance question. The company says the agent works within the permissions already set for each user. Fortune reports that the product is meant to shadow an office worker through the day, offering suggestions and pulling information as events happen. Fortune also quotes Chief Executive Ali Ghodsi describing how he uses the agent himself, though that is one executive's account rather than independent evidence.
What changes in practice, at any size
The first change is about data location. Genie Search covers the Databricks lakehouse and also indexes content in business tools such as Google Drive, Jira and Confluence. According to the press release, teams can index that content where it lives, ingest it into the lakehouse, or query the applications live through MCP, a connection standard for AI tools. A business with its documents spread across several apps would need to decide which of those three routes fits each source.
The second change is about control. Databricks says Genie follows access rules already set in Unity Catalog, its governance layer for data. Administrators control permissions, and the company says every action is auditable. A five-person firm and a group with thousands of staff both need to decide who can grant the agent access, and who reviews the agent's output before it reaches a customer or a payment.
The third change is about cost. The announcement does not state a price for the always-on mode, so the vendor's current pricing is the place to check. An agent that runs continuously uses compute all the time, so the monthly bill depends on how many jobs a business sets up and how often they run.
Limits to weigh before adopting
The product is only described as running inside a Databricks environment. A business that does not use Databricks has no stated route to it from this announcement. The always-on mode is also still in preview or beta, which means its behaviour over long periods has less public track record than the generally available parts. Trade coverage notes that no outside party has yet published results from an agent left running unattended for weeks. Automation of this kind is not worth starting where a wrong action would be costly and nobody is assigned to review the agent's logs.
What to watch next
Three things will show whether the release changes how businesses automate work. The first is whether Genie Search, Ontology Manager and the always-on mode move from preview to general availability, and on what dates. The second is the published pricing for continuous use. The third is whether independent testers publish results from longer runs, since the benchmark figures in the announcement come from the company itself.
How AiStaffo would automate this
For a business that runs its records across Google Drive, Jira and Confluence, AiStaffo would start by mapping which routine tasks, such as follow-ups on open tickets or weekly report compilation, touch those tools and which ones need a person's approval. Where a business already runs Databricks, AiStaffo would connect the relevant sources and set up scheduled jobs that collect, check and draft the routine items. Every output would land in a review queue, so the owner or an operations lead approves anything that changes a record or leaves the company. The owner still decides the rules, the permissions and the approval steps. Book a free automation audit
Questions people ask
Is the always-on mode of Databricks Genie generally available?
Can a business use Genie without an existing Databricks setup?
Which apps can Genie connect to?
How does Databricks control what the agent can see?
How much does the always-on Genie mode cost?
Book a free automation audit
Thirty minutes. We look at one process you run every week and tell you exactly what an AI worker would take off your desk, and what it would not.
