---
type: "post"
id: "databricks-has-introduced-a-new-genie-release-that-48a97e"
slug: "databricks-has-introduced-a-new-genie-release-that-48a97e"
title: "Databricks has introduced a new Genie release that expands the product from conversational analytics into a long-running enterprise agent."
description: "Databricks is expanding Genie from conversational analytics into an autonomous coworker for persistent enterprise work."
retrieval_nugget: "Databricks is expanding Genie from conversational analytics into an autonomous coworker for persistent enterprise work. The release combines scheduled or trigger-driven tasks with shared business context, cross-system search, tool connections and administrative controls—an architecture worth examining for teams deciding how agents should act on company data."
published_at: "2026-11-12"
updated_at: "2026-11-12"
record_date: "2026-11-12"
date_kind: "published_at"
topics: ["enterprise-agents","long-running-agents","agents"]
entities: ["Databricks","Genie","Genie Ontology","Genie Search","Unity Gateway","Unity Catalog","Turing","Genie Agents"]
related_patterns: ["agent-ready-software-exposes-capabilities"]
novelty_label: "new_fact"
what_changed: "Databricks introduced a new Genie release that acts as a continuously working enterprise agent, supported by Genie Ontology and Search and governed through Unity Gateway and Unity Catalog."
prior_refs: []
source_url: "https://www.databricks.com/blog/new-genie-autonomous-ai-complex-enterprise-work"
source_urls: ["https://www.databricks.com/blog/new-genie-autonomous-ai-complex-enterprise-work"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:databricks-has-introduced-a-new-genie-release-that-48a97e"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/databricks-has-introduced-a-new-genie-release-that-48a97e/","markdown":"https://newruntime.com/posts/databricks-has-introduced-a-new-genie-release-that-48a97e.md","json":"https://newruntime.com/posts/databricks-has-introduced-a-new-genie-release-that-48a97e.json"}
markdown_url: "https://newruntime.com/posts/databricks-has-introduced-a-new-genie-release-that-48a97e.md"
---

# Databricks has introduced a new Genie release that expands the product from conversational analytics into a long-running enterprise agent.

## Retrieval answer

Databricks is expanding Genie from conversational analytics into an autonomous coworker for persistent enterprise work. The release combines scheduled or trigger-driven tasks with shared business context, cross-system search, tool connections and administrative controls—an architecture worth examining for teams deciding how agents should act on company data.

## Summary
Databricks is expanding Genie from conversational analytics into an autonomous coworker for persistent enterprise work. The release combines scheduled or trigger-driven tasks with shared business context, cross-system search, tool connections and administrative controls—an architecture worth examining for teams deciding how agents should act on company data.

## What happened
Databricks says its new Genie can take responsibility for long-running work rather than answer only individual questions. Users assign a job, provide access to the tools it needs and set guardrails; Genie tracks progress, follows up when information or approval is missing, and adapts to feedback. Jobs can run on a schedule or respond to triggers such as a metric declining. The company’s examples include monitoring insurance claims for fraud patterns and loss-ratio drift, then routing a case to underwriters, and managing marketing campaigns by investigating underperformance and proposing budget changes for approval. Users can supervise, steer or stop multiple jobs through a conversation and a control panel.

The release centers on shared business context. Genie Ontology connects data with definitions, metrics, relationships and organizational knowledge. Databricks says agents build much of it from existing dashboards, queries and notebooks, while people govern authoritative definitions in Unity Catalog. Personal memory can retain a user’s preferences and instructions. Genie Search combines indexed content from business applications including Google Drive, Jira, Confluence, Workday and SharePoint with live MCP search for fresh updates and applications that are not indexed. Genie can choose among indexed search, MCP and SQL, then return an answer with citations. Databricks says Unity Catalog maps source permissions and checks them before results are returned.

Genie is accessible through Genie One, email, Slack and Microsoft Teams, and Databricks says it is also available in browser, desktop and mobile experiences. Connectors let it read from and act in tools including Slack, Zoom, Microsoft 365, Jira, GitHub, Gmail and Google Drive; custom MCP can connect other systems such as ServiceNow and PagerDuty. Genie can create shareable artifacts such as slide decks, dashboards and interactive HTML pages in conversation. Teams can publish repeatable workflows as skills governed in Unity Catalog, running with the permissions of the user who invokes them. Genie Agents are also described as working across structured data and documents, with content search in Unity Catalog Volumes and page-level citations.

Databricks presents Unity Gateway and Unity Catalog as the control layer. Genie acts with the user's identity and respects Unity Catalog permissions and, for indexed content, source-system permissions. Admins control connectors, MCP tools and approval requirements; users choose what Genie may do autonomously within those limits. The company says every action is traced and auditable, and admins can set workspace and per-user budgets, spending caps and alerts. Databricks reports a benchmark it conducted with Turing: Genie with Ontology 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. The tasks connected information across tables, dashboards and notebooks.

## What the archive adds
New Runtime's earlier coverage described NVIDIA's enterprise agent toolkit as a stack problem: turning company data into useful context while bounding agent actions ([October 22, 2026 note](https://newruntime.com/posts/nvidia-agent-toolkit-expands-grounding-enterprise-agents-in-af91c3/)). Databricks' release is a new product event in that same enterprise-context-and-controls thread, adding a persistent Genie workflow and a described search, ontology and gateway combination.

## Trend signal
This supports the pattern that agent-ready software exposes a small set of observable, permissioned capabilities that external agents can compose safely. Genie connects to applications and MCP tools, while Unity Gateway gives admins connector controls, action traces and spending limits ([Agent-ready software exposes capabilities](https://newruntime.com/patterns/agent-ready-software-exposes-capabilities/)). The announcement describes a product design, not proof that these controls are sufficient in deployment.

## Why now
Databricks frames the change as a response to work that is continuous, complex and data-intensive: assistants can analyze or summarize, but employees still track changes and decide what to do. It says thousands of customers already use Genie in production, providing the context for a move from individual questions toward delegated ongoing responsibilities.

## Why it matters
Persistent agents change the operational question from “Can the model answer?” to “Can the system keep working safely when the relevant facts and permissions live across applications?” Databricks' design separates several concerns that teams should test: business definitions, retrieval across indexed and live sources, identity-aware access, approval gates, and audit and spending controls. The separation matters because a correct answer built on the wrong metric or an unauthorized document is still a failure, and a useful agent that can act indefinitely needs limits beyond prompt instructions.

For builders, the announcement is a practical checklist rather than independent proof of performance. Define which business terms are authoritative; test whether permissions survive search and cross-tool actions; decide which actions require human approval; and set budgets and stop conditions before scheduling work. Measure complete task outcomes and the evidence behind them, not just answer quality or token use. Databricks' reported benchmark suggests shared context may affect both correctness and consumption, but teams should reproduce such tests on their own workflows and compare against alternatives.

## Limits and uncertainty
Databricks' 91% result is a company-reported benchmark built with Turing; the announcement does not establish independent replication or provide enough detail here to assess the full task set and evaluation methodology. The customer examples illustrate proposed uses, not measured outcomes. The announcement does not quantify latency, reliability over long-running jobs, error rates in tool actions or the operational cost of indexing and maintaining context. Teams will need to verify connector coverage, permission behavior and approval controls in their own environments before delegating consequential work.

## Reader takeaway
Genie’s significance is not simply that it can answer business questions. Databricks is packaging ongoing work around business context, cross-system retrieval and controls over identity, tools and cost. Treat the release as an architecture to evaluate: test permissions and task completion on your own workflows before trusting an agent with persistent responsibility.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Enterprise Agents](https://newruntime.com/topics/enterprise-agents.md): Explore the enterprise-agents topic hub.
- topic hub: [Long Running Agents](https://newruntime.com/topics/long-running-agents.md): Explore the long-running-agents topic hub.
- topic hub: [Agents](https://newruntime.com/topics/agents.md): Explore the agents topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)