Databricks Week of Genie graphic reading The New Genie, Autonomous AI for Complex Enterprise Work, with a sample Genie answer and usage chart

Databricks Always-On Genie: Power-User Guide to Jobs, Ontology and Search

by

Last updated: November 12, 2026

Databricks Genie is now an always-on, autonomous agent for enterprise work, Databricks announced on November 11, 2026. You give it a goal and boundaries, and it runs jobs in the background, delegates to subagents, and brings decisions to you through a priority inbox and email. Always-on Genie and Genie Search are entering preview and beta programs.

Contents

What is new in the second-generation Genie

Genie is the family of AI experiences inside Databricks. The Genie documentation lists Genie One for business users, Genie Agents for domain-specific setups, and Genie Code for developers. Until now, you asked Genie a question and it answered.

The November 11 press release changes that model. Genie now runs in autonomous mode on its own compute. It takes ownership of a goal, delegates to subagents, tracks progress, follows up on missing inputs and reaches out when a person must decide.

The table compares the earlier Genie experience with the new release, based on Databricks’ launch blog and press release.

Area Earlier Genie (Genie One) Second-generation Genie
Interaction You ask questions in natural language You assign a goal; Genie keeps the work moving
Execution Answers in a conversation Background jobs on a schedule or on triggers, using its own compute
Parallel work Not described Many jobs at once, with subagents and a control panel for active jobs
Notifications Not described Priority inbox plus email alerts, one email thread per alert
Proactivity Waits for a question Notices repeated work, offers to take it on, and flags issues
Outputs Answers, dashboards, apps Live slide decks, dashboards and interactive HTML artifacts that stay current

Availability and pricing

Availability differs by feature.

  • Generally available: Genie and Genie Ontology, for all Databricks customers.
  • Coming soon to preview and beta programs: the new always-on Genie, Genie Search and Ontology Manager.
  • Pricing model: Databricks says customers pay for what they use, not per seat, and admins can set spending limits by team or user through Unity Gateway.
  • Current promotion: the documentation says Genie One and Genie Agents usage by users is free through January 31, 2027. Service principals are excluded and billed.
  • Genie Code: billed pay-as-you-go since July 8, 2026, with a per-user free monthly allowance.

The announcement did not list per-job or per-token prices for always-on Genie.

How to hand Genie an always-on job

These steps follow the workflow Databricks describes in its blog.

  1. Confirm your workspace has access to the always-on Genie preview or beta, since it is not yet generally available.
  2. Open Genie where you work: Genie One in the browser, desktop, iOS or Android, email, Slack or Microsoft Teams.
  3. Describe the job as a goal, for example “track revenue trends and top customer feature requests and send a weekly brief.”
  4. Give Genie access to the tools it needs, such as MCP connectors for Slack, Jira, GitHub, Microsoft 365 or your calendar.
  5. Set boundaries: decide which actions Genie may take on its own and which need your sign-off, within your company’s Unity Gateway rules.
  6. Choose a schedule or a trigger, such as when a metric dips or when more than 10 customers ask for the same feature.
  7. Monitor the job from the priority inbox and the control panel for active jobs. Steer, approve or stop jobs in one conversation.

Genie keeps personal memory of your preferences, so you need not repeat them for each job.

Genie Ontology is the business context behind every answer. According to the blog, agents build most of it automatically from dashboards, queries and notebooks. Teams set authoritative definitions in Unity Catalog, and the OntoRank algorithm gives those precedence. External BI sources such as Power BI and an organizational directory connector, starting with Workday, extend it.

Ontology Manager lets admins monitor context health. Proactive gap alerts flag frequently asked-about concepts that lack governed definitions.

Genie Search combines an index of content from apps such as Google Drive, SharePoint, Jira, Confluence and Workday with live search through MCP. Databricks says it now offers more than 100 connectors. Genie Search enforces each source’s access controls before returning results, so users see only what they could open in the original app.

On accuracy, Databricks cites a benchmark built with Turing in which Genie answered correctly in 91% of task runs, 1.6 times the rate of the stronger of two frontier agents, while using less than half as many tokens.

Power-user workflows

Daily briefing. Ask Genie to monitor your email and Slack, prepare you for the day and alert you as new items arrive.

Metric watch. Databricks gives examples such as monitoring insurance claims for fraud patterns and loss-ratio drift, or reviewing campaign performance and proposing weekly budget changes for approval.

Incident follow-through. With custom MCP connections to tools such as ServiceNow or PagerDuty, Genie can investigate an incident using GitHub and Jira changes, keep the ticket current and update the team.

Reusable skills. Publish a repeatable workflow as a skill. Skills live in Unity Catalog and run with the permissions of whoever uses them.

Use Genie from other assistants. Genie plugins bring its context into ChatGPT, Claude, Copilot and Grok Bot; see our Grok Bot 1.0 guide and Claude interactive connectors guide for how those assistants handle connected apps.

If you are comparing long-running agent platforms, our guides to Mistral Managed Agents, Perplexity Teammates and the ChatGPT Business Hub cover similar ideas from other vendors.

Governance and cost controls

  • Genie acts with your identity and accesses only data you are authorized to see in Unity Catalog.
  • When you need more access, Genie can prepare an access request for you to review; data owners keep approval authority.
  • Admins control connector access, MCP tools and approval requirements through Unity Gateway.
  • Every action is traced and auditable. Admins can set workspace and per-user budgets, spending caps and alerts.

Limitations

  • Always-on Genie, Genie Search and Ontology Manager are preview or beta features, so behavior may change.
  • Genie runs inside a Databricks environment; it is not a standalone consumer assistant.
  • Analysts quoted by TechTarget said enterprises still need stronger controls over what agents do in production, and that Genie should signal when it is unsure of an answer.
  • Accuracy figures come from a Databricks benchmark and may not reflect your data.

Frequently asked questions

What is the new Databricks Genie?

It is the second generation of Genie, announced on November 11, 2026. Databricks describes it as an always-on, autonomous agent for enterprise work: you give it a goal, and it delegates to subagents, tracks progress, follows up on missing inputs and contacts you through a priority inbox and email when a decision is needed.

Is always-on Genie available now?

Not yet for everyone. Databricks says Genie and Genie Ontology are generally available to all customers, while Genie Search, Ontology Manager and the new always-on Genie are coming soon to preview and beta programs. No date was given for general availability of those features.

How much does Databricks Genie cost?

Databricks says Genie is part of its Data + AI Platform and customers pay for what they use, not per seat. Its documentation says Genie One and Genie Agents usage by users is free through January 31, 2027, excluding service principals. Genie Code moved to pay-as-you-go billing with a per-user free monthly allowance on July 8, 2026.

What is Genie Ontology?

Genie Ontology is a layer of business context: definitions, metrics, relationships, people and processes. Agents build most of it automatically from existing dashboards, queries and notebooks, and teams set authoritative definitions in Unity Catalog. An algorithm called OntoRank gives those governed definitions precedence.

Can I use Genie from ChatGPT or Claude?

Yes, according to Databricks. Genie plugins and Genie MCP let tools such as ChatGPT, Claude, Microsoft Copilot Cowork and Grok Bot use Genie Ontology and Genie Search with the same governed context and permissions.

Sources

Related answers

Leave a Comment