Atlan MCP server: Claude builds a glossary from Confluence

Atlan presenter

0:00 So let's talk about how we're going further by introducing our MCP server. Okay, switch screens for y'all. Perfect. During my demo, I'm gonna be using Claude's, but I think we first need to level set with what the heck is an MCP server? I think of Claude like a really smart assistant. On its own, it knows a lot, but doesn't automatically know how to work with your systems like Atlan. An MCP server is the universal adapter that connects Claude to those systems. That means instead of me manually hunting for information or updating assets in Atlan, I can just ask my assistant to do it. The MCP server makes Claude talk to Atlan safely and reliably.

0:49 In short, MCP servers turn AI from just answering questions into a true agent that helps me be more productive. Let's hop in and see how this can help us. We have a Confluence space called the Customer Value Handbook here, and in there, we have a lot of unstructured information spread across multiple pages, so I'd like to turn this into a glossary in Atlan. I ask Claude, "Hey, can you go to Confluence and summarize what you found?" Claude starts searching vigorously to gain the context that it needs to build the Atlan glossary. Then it summarizes the results, so I know exactly what it found. You know what? I want to use this to build the glossary.

1:32 So next, I do a little prompt engineering. I ask it to create categories and terms in the glossary, Customer Success Handbook. It lists out all the steps that it'll take to perform these actions, then asks, "Hey, are you good with this?" Which I gladly say yes as I eat my next donut. Funny enough, I forgot to create the glossary for this demo. Claude sees that, and it decided to create it for me, already correcting my mistake. I then go on to form all of the categories and terms. Once done, it provides a wonderful summary of what happened.

2:07 We see it created forty-five different terms and various categories for me, but the real test is, let's check Atlan to see if it worked. Right there in the top left, I see my customer success glossary and all of the categories and terms underneath, each with a description and a draft status certificate as I told Claude to do. With an MCP server, documenting your data is no longer a chore. It's as simple as asking Claude. And this is just the beginning. For example, you could ask the AI assistant to spin up an industry-specific glossary template if you're just starting your governance journey, or have it draft documentation for a brand-new database.

2:49 Ultimately, we're most excited to see the creative use cases our customers come up with. Atlan's MCP server unlocks a whole new way to work, turning an AI assistant or agent into a true teammate that helps you get real work done faster. PK, turn it back over to you.

Prukalpa Sankar

3:10 Ever since we launched our MCP server, we have been blown away by the use cases that the community has imagined with us, from engineers bringing Atlan's MCP into wide coding environments like Cursor, to analysts and stewards writing up documentation, to exposing context into agents via agent orchestration frameworks to do automated compliance reporting. We think that we've just scratched the surface of what's possible, and we are super, super excited to imagine this future with you.