The new Genie: autonomous AI for complex enterprise work

Databricks introduced a new version of Genie, an autonomous AI coworker that proactively handles long-running enterprise tasks, grounded in business context and operating within organization-set guardrails. The release adds Genie Ontology, an agentic ontology built by agents and governed by people, Genie Search for finding information across a lakehouse and connected business systems, and Unity Gateway, which lets admins control tool access, trace actions, and set spending limits. Databricks said thousands of customers already use Genie in production, including Zillow, Unilever, adidas, Atlassian, and AstraZeneca, and users can work with Genie through Genie One, email, Slack, and Microsoft Teams.

Genie is always on, proactively working alongside you on long-running enterprise tasks, grounded in your business context, within your organization’s guardrails. by Cynthya Peranandam https://www.databricks.com/blog/author/cynthya-peranandam and Elise Georis https://www.databricks.com/blog/author/elise-georis • Genie is now an autonomous AI coworker that handles long-running tasks on your behalf, operating within the guardrails you set. • Genie Ontology, an agentic ontology built by agents and governed by people, provides trusted business context, while Genie Search finds information across your lakehouse and business systems to help Genie deliver accurate and complete answers. • Unity Gateway lets admins control tool access, trace actions, and set spending limits to keep autonomy under control. AI assistants can run analyses and summarize meetings, but employees still manage the work around them. That’s because most enterprise work is continuous, complex, and data-intensive. To monitor pipeline performance or manage budgets, someone has to track changes, decide what matters, and take action. Earlier this year, we introduced Genie One to help teams understand their data and act on it. Thousands of customers use Genie in production, including Zillow https://www.databricks.com/customers/zillow/lakebase , Unilever https://www.databricks.com/customers/unilever/genie , adidas https://lnkd.in/p/gpN87b6b , Atlassian https://www.databricks.com/customers/atlassian/security-lakehouse , and AstraZeneca https://www.databricks.com/blog/bringing-visualizations-life-multi-agent-systems-vega-lite . Today, we’re introducing the new Genie: an autonomous AI coworker that proactively handles long-running enterprise work for you. Genie stays on the job continuously, tracking progress, following up on missing inputs, and taking action as new information arrives. You set the guardrails, supervise its work, and approve where needed, and Genie takes care of the rest. That autonomy depends on business context and clear controls. Genie Ontology gives Genie a map of your business, including its definitions, metrics, relationships, and organizational knowledge. Genie Search helps people and agents find information across your lakehouse and connected business applications. Unity Gateway and Unity Catalog govern Genie within your existing permissions and policies, while giving admins visibility into its actions, usage, and cost. You can work with Genie https://www.databricks.com/product/genie through Genie One, email, Slack, and Microsoft Teams. Give it a job and access to the tools it needs, and it takes responsibility for ongoing work. It checks in when it needs input or approval, and it adapts to your feedback. For example, Genie can monitor incoming insurance claims for emerging fraud patterns and loss-ratio drift, then assemble the case file and route it to underwriters when a segment moves outside guidelines. In marketing, Genie can manage campaigns around the clock, investigate any underperformance, and propose weekly budget changes for approval. It can also monitor your email and Slack, prepare you for the day, and alert you as new things come up. What makes this possible is context from Genie Ontology, which tells Genie how your business works—from what counts as an active customer to which revenue metric is authoritative. Personal memory retains your preferences and instructions, so you don’t have to repeat them for every job. Genie is also proactive. It notices work you repeat, like a weekly update, and offers to take it on. It can also flag issues that need your attention even before you set up a job. Once a job is running, it can run on a schedule or respond to triggers, such as when a metric dips or when more than 10 customers ask for the same feature. Genie can work on many jobs at once. You can supervise, steer, and stop them in one conversation, with a control panel for active jobs and outputs. You decide what Genie does on its own and what needs your sign-off, within your company guardrails. We're also expanding Genie https://www.databricks.com/blog/genie-one-ai-cowork-home-everyone with new ways to find information, create finished work, and collaborate across tools. One place to find answers: Search employees, documents, tickets, and more from the same chat bar where you ask questions. Genie Search brings results from sources like Google Drive, SharePoint, Jira, Confluence, and Workday together with Databricks dashboards, tables, Databricks Apps, and Genie Agents. Organizational directories help you explore team structures and find the right colleague. Artifacts ready to share: Describe the slide deck, dashboard, or interactive HTML page you need, and Genie builds it directly in the conversation. Refine it in chat, then present, share, or reuse it. Scheduled tasks keep each artifact current as your data changes, while shared skills apply company brand guidelines. "With artifacts in Genie, we have full creative control over how we present our data. We can get the exact look and feel we want while keeping everything connected to our governed data — it's the best of both worlds."- Guilherme Vivian, Data Manager, Agibank Work across applications: MCP connectors let Genie read from and take action in popular tools like Slack, Zoom, Microsoft 365, Jira, GitHub, and your calendar, all grounded in the same ontology that gives Genie the definitions, relationships, and context to understand your business. Through custom MCP you can also connect to ServiceNow, PagerDuty and more: Genie can investigate an incident using GitHub and Jira changes, keep the ticket fresh, and keep the team updated until the issue is resolved. And when someone gets great results from a repeatable workflow, they can publish it as a skill and share it with their team. Skills live in Unity Catalog and run with the permissions of whoever uses them. Genie wherever you work: Message or @mention Genie in Slack and Microsoft Teams, or use it in your browser, on desktop, and on iOS and Android. Plugins also bring Genie’s business context into AI tools such as ChatGPT and Claude. Genie works alongside every business user, and Genie Agents https://www.databricks.com/blog/enterprise-context-action-genie-agents-now-analyze-more-data-sources-transform-your-work bring specialized expertise to entire teams. Thousands of teams already use them in marketing, sales, finance, and operations. A department configures an agent with the data, files, tools, and instructions it needs, then manages its access and use. Teams can work with agents in Genie One or through APIs, custom applications, and automated workflows. Work across structured and unstructured data: Genie Agents now works across not only structured tables but also PDFs, slides, and more. Content search in Unity Catalog Volumes makes this possible without a separate vector index. A sales leader asking why a region is losing deals can get an analysis that combines pipeline data and win-loss reports, with page-level citations that make the supporting evidence easy to inspect. From investigation to action: Connectors let Genie Agents read from and act in tools including Slack, Atlassian, GitHub, Gmail, Google Drive, and Microsoft 365. An agent can investigate an issue, create a ticket, and prepare a summary for the team's channel, carrying the findings into the systems where teams work. Share expertise beyond your organization: OpenSharing for Genie Agents lets teams share governed agents with partners and customers, pre-loaded with the data, instructions, and knowledge they need to get value right away. For more sensitive use cases, Genie Clean Rooms lets recipients ask questions and get answers without exposing the underlying data or proprietary logic, with controls such as conversation and output limits. "Genie Agents let our domain experts productize their knowledge of the data directly. What used to take a full development cycle now takes days, and every answer stays inside our governance boundary."- Priyanka John, Vice President, S&P Global Energy Whether Genie manages a single task or an ongoing responsibility, it needs to understand your business and how information across systems connects. Genie Ontology https://www.databricks.com/blog/announcing-ga-genie-ontology gives AI across your company that shared understanding. It connects structured and unstructured data with definitions, relationships, and organizational knowledge. Built by agents, governed by people: Agents build most of Genie Ontology automatically from the dashboards, queries, and notebooks people already use, and it keeps learning with every query and AI interaction across your organization. External knowledge sources extend that learning to business logic in BI tools such as Power BI. Teams establish authoritative definitions in Unity Catalog, and the OntoRank algorithm gives those definitions precedence, so every agent works from the same governed understanding of the business. Understand people and teams: Organizational directories, starting with Workday, bring people, teams, and reporting relationships into Genie Ontology. OntoRank combines those relationships with activity signals that show who actually works together, so Genie can weigh what's most relevant to the person asking. Retain admin controls: Ontology Manager gives teams a shared place to manage and customize context as the business evolves. Admins can monitor context health, while domain experts curate the knowledge they know best. Unity Catalog Domains , Pages , and Ontology Snippets support both expert-curated definitions and the knowledge Genie learns automatically. Proactive gap alerts flag frequently asked-about concepts that lack governed definitions, and owners can promote useful learned knowledge into authoritative Pages. Shared business context also helps improve accuracy and reduce token use. In a new benchmark built with Turing, we compared Genie, powered by Genie Ontology, with two leading frontier agents on the same tasks in the same Databricks workspace. Genie answered correctly in 91% of task runs, 1.6 times the rate of the stronger frontier agent, while using less than half as many tokens. The tasks required connecting knowledge spread across tables, dashboards, and notebooks, not simply accessing data. "Genie Ontology is exactly what we need to scale AI without losing the business context that makes it useful. It's a force multiplier for our data teams." - Noah Cepela, SVP, Analytics and Insights, Mode Global Genie Ontology gives Genie a shared understanding of your business. To put that understanding to work, Genie also needs to find relevant information wherever it lives, from lakehouse tables to documents, tickets, and wikis. Just as importantly, it needs to enforce each asset’s access controls. Today, we're introducing Genie Search , a new kind of hybrid AI search that brings enterprise search into Databricks, with no separate search infrastructure to procure and no separate permission model to maintain. Genie Search combines an index of content from core business apps like Google Drive, Jira, Confluence, Workday, and SharePoint with live search through MCP. The index provides fast, accurate access, while MCP retrieves the most recent updates and the long tail of apps that can't be indexed. For each task, Genie draws from the index, MCP, or SQL, then composes one answer with citations. The result is more complete answers with fewer tool calls and tokens. Beyond the standard benefits of a search index, Genie Search also feeds Genie Ontology, so your tickets and documents become evidence OntoRank weighs when deciding which sources to trust. Through Unity Catalog, Genie Search maps each source's access controls alongside its content and enforces them before returning any results, so nobody sees a document in Genie that they couldn't open in the original application. The same business context that powers Genie can also support work in other AI assistants. Connect your data once, let Genie build your ontology from how your teams work, then bring that context through Genie plugins https://www.databricks.com/blog/bring-your-business-context-chatgpt-claude-microsoft-copilot-cowork-and-grok-bot to ChatGPT, Claude, Copilot, and Grok Bot. You can also use Genie MCP https://www.databricks.com/blog/genie-one-mcp-now-generally-available directly to connect Genie to other AI tools. Agents can hand an entire business question to Genie or request the context needed for a larger task. Genie applies governed definitions, respects user permissions, and cites its sources. As Genie takes on ongoing work across enterprise systems, admins need to control who uses it, what it can access, and how much it costs. Unity Gateway https://www.databricks.com/product/artificial-intelligence/unity-gateway brings those controls together, letting admins govern Genie alongside coding agents, models, MCP tools, and skills. Data access: Genie acts with your identity and accesses only information you’re authorized to see. It respects Unity Catalog permissions and, for content indexed by Genie Search, each source system’s permissions. When users need additional access, Genie can prepare an access request for them to review and send. Admins and data owners retain approval authority. Tools and actions: Admins control connector access, MCP tools, and approval requirements. Business users choose what Genie can do autonomously within those limits. Unity Catalog governs reusable skills, and the same controls apply when external agents such as Claude call Genie through MCP. Usage, cost, and auditing: Every action Genie takes is traced and auditable. Admins can set workspace and per-user budgets, spending caps, and alerts. They can also ask Genie to investigate usage and agent traces for wasted tokens and costly tool calls. With Genie, teams can move from asking AI for help with individual tasks to giving it responsibility for ongoing work, with shared business context and controls built in. Learn more: Join us for Data + AI Live: Context Is Just the Beginning , with Ali Ghodsi and Greg Brockman of OpenAI. Watch live or catch the replay. https://www.databricks.com/resources/webinar/context-is-just-the-beginning Subscribe to our blog and get the latest posts delivered to your inbox.