Databricks Unveils Genie Plugins for Four AI Tools

On November 11, 2026, Databricks announced the launch of native plugins for its Genie platform, specifically designed for four prominent AI tools: ChatGPT, Claude, Microsoft Copilot Cowork, and Grok Bot. These new integrations enable each agent to securely access business context stored in Databricks, streamlining their operations and offering them a richer set of functionalities.

Enhancing AI Agent Capabilities

Prior to this launch, Genie had already enabled connections to various external assistants. A prior announcement on September 22, 2026, revealed the framework for an integrated Model Context Protocol (MCP) that facilitated communication between AI agents and external resources. The latest plugins provide direct interfaces within the specified tools while maintaining the existing context layer established by Genie.

A Deep Dive into Genie Ontology

At the core of Genie’s functionality is Genie Ontology, which combines clearly defined metrics and established business relationships with context derived from various enterprise elements, including governed tables, dashboards, queries, and notebooks. By assessing the relevance and authority of this material, Genie applies appropriate permissions, ensuring users only access necessary information. This design allows organizations to maintain a consistent definition of metrics, which can be universally applied across all connected AI agents.

When posed with a question, Genie translates business terminology, identifies data accessible to the user, generates and runs SQL queries, and presents answers supported by citations from Databricks sources. Genie can also delegate specific inquiries to dedicated Genie Agents tailored to different fields—like finance and marketing—to extract meaningful insights from even unstructured enterprise data.

AI Agent Workflow Examples

The practical applications of these plugins have already demonstrated distinct workflows across various products. For example, a ChatGPT Dot operates continuously and can evaluate feature adoption against defined benchmarks, compile relevant support topics, and summarize findings for distribution via Slack. Meanwhile, Claude is capable of analyzing regional revenue downturns using verified financial metrics and relevant documents, producing an evidence-based report complete with citations.

Microsoft Copilot Cowork utilizes insights from Azure Databricks to create professional briefs in Word, generate visual data in Excel, and assemble client presentations, awaiting user approval before any distribution. Lastly, a Grok Bot involved in growth analysis can assess campaign performance based on governed definitions, recommend budget reallocations with substantiated evidence, and prepare retargeting strategies, all pending approval from a designated growth lead. In each of these scenarios, Genie provides authoritative business metrics while the agents manage documentation and execution of tasks.

Security and User Permission Management

To ensure data integrity and security, permissions provided by Unity Catalog are enforced across all user requests, thereby restricting results to only what is permissible for each user. The Genie One MCP's documentation suggests implementing OAuth authentication to carry the identity of the end user into requests, while automated tasks can also utilize service-principal authentication. The process of enabling these plugins starts with integrating them into the agent workspace. The plugins for ChatGPT and Claude leverage default OAuth applications, allowing users to log in using their Databricks credentials without needing administrative setup.

Since Genie’s responses operate asynchronously, the system requires time to search through data and execute SQL queries. Tools designed for this purpose help to manage query results effectively, particularly for larger datasets, as they ensure that responses remain within manageable limits.

Looking Ahead: The Future of Genie Plugins

The existing beta MCP endpoint will become obsolete after November 30, 2026, necessitating a transition to the Genie One MCP endpoint for continued access to these resources. Additionally, the quality of information provided by Genie is contingent on the underlying data framework. The implementation guide from September 1, 2026, emphasizes that while Genie Ontology is functional from day one, optimal performance relies on quality data management practices, including consistent entity records and thorough governance.

With this latest rollout, Databricks establishes a strong governance model across different AI agents. As these tools evolve from analysis to actionable outcomes, the uniform ontology ensures that they utilize consistent data, enabling users to access enterprise insights tailored to their permission levels.