Genie One MCP Guide: Connect Governed Data to Any Agent
Read the Portuguese translation →
Author: Databricks · Watch on YouTube
Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.
Original language: English
Original video: Genie One MCP Guide: Connect Governed Data to Any Agent · Channel: Databricks · 03:11 · Watch on YouTube
Topics: AI agents, Data, databases and embeddings
Summary
Section titled “Summary”Ben, a Databricks product manager, demonstrates how Genie One MCP brings governed business context into an agent workflow in Cursor. He asks about the difference between gross booking value and collected revenue, and Cursor uses Genie One MCP to search the Genie ontology for relevant glossary entries. The cited definitions appear in Cursor, where Ben can use them to ask for a gold layer table comparing the two measures. The demo shows how Genie context can help ground pipeline code in the company’s business semantics without leaving the development workflow.
Key points
Section titled “Key points”- Genie One MCP lets users leverage Genie One as a trusted data co-worker within other agent workflows.
- The demo uses Cursor while Ben develops a data pipeline.
- Ben asks about the difference between gross booking value and collected revenue.
- Cursor uses Genie One MCP to search the Genie ontology.
- Genie identifies relevant glossary pages and provides cited definitions.
- The response is visible in Cursor, so Ben does not need to open the Genie UI.
- Ben uses the context in a follow-up request to create a gold layer table comparing the measures.
- The resulting SQL is grounded in the company’s business semantics and proprietary insights.
Chapters
Section titled “Chapters”| Start | Chapter | Summary |
|---|---|---|
| 00:00 | Introducing Genie One MCP | Ben introduces Genie One MCP and explains how it can bring Genie One into other agent workflows. |
| 01:02 | Searching the Genie ontology | Cursor uses Genie One MCP to search the ontology for glossary information about two revenue measures. |
| 02:06 | Using business context in Cursor | Ben sees the cited definitions in Cursor and asks the agent to compare the measures in a gold layer table. |
| 02:37 | Generating grounded SQL | Cursor uses the context from Genie to start writing SQL grounded in the company’s business semantics. |
Quotes
Section titled “Quotes”The Genie One MCP allows you to leverage Genie One as your trusted data co-worker — 00:00
it’s searching the Genie ontology — 01:02
I never really needed to go into the Genie 1 UI. — 02:06
the code I’m writing is actually grounded in what our business semantics — 03:11
Concepts and entities
Section titled “Concepts and entities”Genie One MCP, Genie One, Cursor, Databricks, Genie ontology, gold layer table
Full transcript (original language)
Section titled “Full transcript (original language)”00:00 [music] >> Hi, I’m Ben and I’m a product manager here at Databricks. Um and I’m here to demo uh what I think is a super cool feature of our Genie One product, which is the Genie One MCP. The Genie One MCP allows you to leverage Genie One as your trusted data co-worker um in whatever other agent workflow you may be using. So, for today’s example um I have Cursor open and I’m working on developing a
00:31 data pipeline um and I’m new to this team. There are some semantic information that I don’t really know, but I’ve been tasked with providing a gold layer table comparing two different ways that we might measure revenue. So first what I’m going to do is go into my agent, which I’ve been using to develop this pipeline itself and just ask quickly um for Wonder Bricks which is the company I’m working for what’s the difference
01:02 between gross booking value and collected revenue? And as you’ll see Cursor will use the Genie One MCP. You can see our little Databricks logo here. Um and if you click in you can actually follow along in our MCP apps view. So, you’ll see that it’s searching the Genie ontology, which is awesome. Um but I can actually just go ahead and click into Genie itself and follow along um
01:34 for an even more expanded view. And you can see that it’s it’s searching here and we’ll just wait for a moment for it to gather the right context um and then provide me with the response I need. So, here you can see that in its search results it was actually able to identify some relevant glossary pages from our Genie ontology. Um, and now it’s formulating its final response. Um, and you can see exactly
02:06 the differences here in these definitions and the two cited, um, glossary entries here. And so now I’m going to go back into Cursor. And you can see this full response is actually reflected here in line. So I never really needed to go into the Genie 1 UI. Um, but then when I scroll down, um, I now have this context integrated into my same workflow where I’m developing this pipeline. So now I can ask a follow-up question basically to help continue this workflow. So I can say add
02:37 a gold layer table comparing the two. And now what Cursor is going to be able to do is leverage that context that Databricks 1 just gave it. And so what Cursor can do now is leverage that context that it got from Genie, um, parse out that response and then help me to develop this actual gold layer table. As you can see, it’s starting to actually write out the final, uh, SQL here. Um, and this just allows
03:11 me to ensure that the code I’m writing is actually grounded in what our business semantics, uh, actually mean. So I’m doing good data work that’s grounded in our company’s, um, proprietary insights. And that’s all thanks to Genie. >> [music]