# Databricks launches Genie plugins for ChatGPT, Claude, Copilot Cowork and Grok Bot

published: 2026-11-11T15:05:40.175Z
topic: mcp
source: https://www.databricks.com/blog/bring-your-business-context-chatgpt-claude-microsoft-copilot-cowork-and-grok-bot
url: https://aiagentnews.io/stories/databricks-launches-genie-plugins-for-chatgpt-claude-copilot-cowork-and-grok-bot

Databricks has released Genie plugins that let agents in ChatGPT, Claude, Microsoft Copilot Cowork and Grok Bot query a company's governed business context. Agents without a native plugin can reach Genie through the Genie MCP server.

Databricks has introduced Genie plugins that give agents in ChatGPT, Claude, Microsoft Copilot Cowork and Grok Bot access to a company's governed business context. Genie uses Genie Ontology for definitions and relationships, then writes and runs SQL over data the user is permitted to see. It returns answers with citations to its sources.

For specialized questions, Genie can route work to domain-specific Genie Agents, such as a marketing agent for pipeline attribution or a finance agent for operational costs. Genie also draws insights from large volumes of unstructured data. In clients that support the MCP Apps extension, answers can include interactive visualizations and Ontology citations in the conversation.

An agent can either hand a business question to Genie or keep its own task and ask Genie only for the context it is missing. Databricks gave examples of a product dot in ChatGPT tracking feature adoption, a Claude analysis of a revenue miss, a Cowork account-planning workflow that waits for approval before sending, and a Grok Bot growth teammate that runs campaign reviews on a schedule.

## Why it matters if you build agents

Agents can pull certified metric definitions from Genie instead of searching tables and documents on their own, which reduces guessing about where answers live. Builders should confirm that the permissions Genie applies match what each agent user is allowed to see. The Cowork and Grok Bot examples also suggest planning for human approval before an agent sends messages or changes campaigns.