Databricks Plugs Genie Business Context Into ChatGPT, Claude, Copilot and Grok

Databricks launched native Genie plugins on November 11, 2026, for ChatGPT, Claude, Microsoft Copilot Cowork and Grok Bot. The move gives all four agents governed access to business context held in Databricks, while other compatible agents can still connect through the Genie One MCP server.

This isn't Databricks' first step into agent territory. On September 22, 2026, the company announced Genie One MCP, which described governed context for assistants and coding agents over Model Context Protocol — the emerging standard for connecting AI agents to external tools. The new plugins sit a layer above that: they add native interfaces inside the named products while using the same context infrastructure underneath.

The stakes here are straightforward. Enterprises want AI agents that answer questions with the company's own approved numbers, not guesses. Whoever supplies that governed context layer sits in a powerful position as agents move from chat to taking real actions.

How does Genie actually supply business context?

The mechanism centers on what Databricks calls Genie Ontology. It combines deliberately defined metrics and business relationships with context inferred from governed tables, queries, dashboards, notebooks and other enterprise assets. Genie then ranks that material by authority and relevance, applies permissions, and supplies the definitions, relationships and source context it uses to interpret a question.

The practical payoff: an organization can maintain an approved metric definition once and apply it across every connected agent. No more reconciling three different definitions of "active user" depending on which assistant you asked.

When Genie receives a question, it runs a defined pipeline:

• It resolves business terms against the ontology
• It finds data the user is allowed to access
• It writes and runs SQL
• It returns an answer with citations to Databricks sources

Genie can also delegate specialized work to domain-specific Genie Agents — marketing for pipeline attribution, finance for operational costs — and can draw insights from unstructured enterprise data.

How much control does each agent keep?

The integration is flexible about where the division of labor sits. An agent can hand Genie an entire business question, or request only the context it's missing before continuing with its own tools — editing files, drafting plans or posting to Slack.

Clients that support the MCP Apps extension get the richer experience: they can display Genie's progress, visualizations, results and ontology citations in an interactive panel. Other clients continue to receive plain text results.

In short: Genie supplies the governed business numbers, and the agent handles documents, communications or campaign execution.

What do the workflows look like in practice?

Databricks' product examples show how the same context layer flexes to different agents.

A ChatGPT Dot — an always-on agent with its own computer — can monitor feature adoption against targets using a certified active-user definition, pull related support themes and post a summary to Slack.

Claude can investigate a regional revenue shortfall using certified finance metrics, pricing memos, sales-call notes and competitor reports, then produce an analysis with citations behind each claim.

Microsoft Copilot Cowork can use account and usage trends from Azure Databricks to populate Word briefs, Excel charts and client presentations, then wait for approval before sending anything.

A Grok Bot growth teammate can review campaigns using governed definitions for campaign spend, qualified opportunities and pipeline, recommend budget shifts with evidence, and prepare retargeting or lifecycle campaigns after a growth lead approves them.

The common thread is approval gates before anything goes out the door — agents act, but humans sign off on actions with real consequences.

What about permissions and security?

Unity Catalog permissions are enforced on every request, so results stay limited to what each user may access. That enforcement travels with the question, not the agent.

The Genie One MCP documentation recommends on-behalf-of user OAuth authentication, which carries the end user's identity into the request. Service-principal authentication is also supported for automated workloads.

Setup is deliberately light. It starts by enabling the plugin in the agent workspace. The ChatGPT and Claude plugins use default OAuth apps, so users can sign in with their Databricks accounts without an administrator having to create an OAuth application.

What are the operational limits?

Genie responses are asynchronous because the system searches data and executes SQL — answers aren't instant. Standard response tools truncate query results to protect the model's context window, while a separate tool retrieves the full schema and rows. Very large results can still be truncated even with that tool.

There's also a deadline worth marking. The older beta MCP endpoint reaches end of life on November 30, 2026. Requests to it will fail after that date, and affected workloads must move to the generally available Genie One MCP endpoint.

Answer quality ultimately depends on the data estate behind the ontology. Databricks' September 1, 2026 implementation guide says Genie Ontology works on day one, but higher accuracy requires clean data, consistent entity records, useful metadata, governed metric definitions, certified assets and continuing evaluation. The guide recommends starting with one domain and strengthening its definitions, sources, governance and evaluation before expanding.

The takeaway: the plugins work out of the box, but garbage in still means garbage out — governed AI answers are only as good as the governed data underneath them.

Why does one context layer across four agents matter?

The launch gives Databricks a consistent governance layer across agents with very different native strengths. ChatGPT, Claude, Copilot Cowork and Grok Bot each bring their own document, communication and execution capabilities; Genie brings the single source of truth for business numbers.

That positioning gets more valuable as agents mature. As those agents move from cited analysis to approved actions, the same ontology can determine which numbers they use and which underlying enterprise data each person is allowed to see — making Databricks the arbiter of what every agent in the enterprise "knows."