On November 11 Databricks released Genie plugins, which it says work natively in ChatGPT, Claude, Microsoft Copilot Cowork and Grok Bot; agents without a native plugin can connect through the Genie One MCP server. An agent can hand a business question to Genie, which interprets terms using certified definitions in Genie Ontology, writes and runs SQL only over data the user may access, routes to domain Genie Agents such as finance or marketing when needed, and returns an answer with citations. Or the agent can just fetch the missing business context and keep working with its own tools.
Key Takeaways
- ✓Native plugins in 4 clients: ChatGPT, Claude, Microsoft Copilot Cowork and Grok Bot; other agents connect through Genie One MCP (system.ai.genie_one_mcp)
- ✓Governance: SQL runs only over data the user may access, metrics use certified Genie Ontology definitions, and answers cite their sources
- ✓Legacy endpoint: the Beta /api/2.0/mcp/genie endpoint reaches end of life on 2026-11-30; tool names and arguments are unchanged, so migrate to the GA server
- ✓MCP Apps: supporting clients show interactive charts and Ontology citations inside the conversation
- ✓Benchmarks: no accuracy or latency figures published
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background and pain points
When agents answer business questions, the bigger risk is often using the wrong definition rather than missing data: "active user" can mean different things in different tables. Databricks says frontier agents facing enterprise data often take a random walk across MCP servers, documents and tables. Genie plugins aim to give agents a map of the business first.
Architecture and mechanisms
The core is Genie Ontology: governed metric definitions, table relationships and learned organizational knowledge, maintained once in Databricks and served to agents through plugins or Genie One MCP. Agents can hand a whole question to Genie, which interprets terms, writes and runs SQL within the user's permissions, routes to domain Genie Agents if needed and returns a cited answer; or they can keep the task and pull only missing context. Genie One MCP is a Databricks-managed MCP governed through Unity Gateway; OAuth clients request the ai-gateway scope.
Benchmarks
Databricks has published no accuracy, latency or cost comparisons, only usage scenarios in ChatGPT, Claude, Copilot Cowork and other clients. Validate with your own data and definitions.
Getting started
Enable the Genie plugin in your agent workspace, or point clients such as Claude Code or Cursor at your workspace Genie One MCP URL and sign in with the default OAuth app. If you still use the Beta endpoint, migrate to system.ai.genie_one_mcp before November 30; tool names and arguments stay the same. Certify core metrics in Genie Ontology first.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.