MCP server

You've done your research, but now you have to analyze your results. Three surveys, a test, and maybe an interview round on top. And you have to turn those results into something that your team can actually use to help drive your product forward. That's what Lyssna's MCP Server is built for. I'm Cal, senior product manager at Lyssna, and today I'll show you what the Lyssna MCP Server is, how to set it up, and a few ways you can use it to analyze your research in the tools that you're already working in.

Lyssna's MCP Server connects your research data to the AI assistant that you already use, Claude, ChatGPT, Microsoft Copilot, Gemini, or Cursor. That means you can ask a question and get an answer grounded in what your research participants actually said. It covers usability tests, surveys, and interviews.

Now you can find research and participant responses by describing what it was about, not just the exact study name. It's a read-only connector, meaning it can answer questions based on your real study data, but it can't yet create studies or write anything back into Lyssna. It's now generally available on every paid plan with no wait list, and it's built to connect to any other MCP client via URL, even those that I haven't mentioned, to support your internal agents and other AI use cases.

Setting this up takes about a minute. Head to settings, and in your profile section, scroll down to the Lyssna MCP area. That's where you'll be able to connect different AI tools easily to your Lyssna data. Click through and install the plugin or follow the instructions for other models. Then all you need to do is allow access to your Lyssna account, and you'll be ready to start.

Asking your assistant about your Lyssna studies.

And that's it. Now, if your company has an enterprise account with your AI platform, things are a bit different. An admin will set up the MCP once for the org, and then you'll sign in under your own account. You can find more setup instructions in our help center article. If you're an admin and you want to turn off MCP for the whole account, that toggle lives within feature access in settings.

Here's the one that we hear about the most. So you've run a few studies over the past few months, and you need to analyze everything that you've learned to share with the team. But you're dreading reopening and analyzing the studies one by one. Now, you can ask something like, "What have participants said across the last three months of our Lyssna studies about our checkout?"

Now, what comes back is pulled from what your participants actually said across all the relevant studies at once. You don't need to know the exact study name. Lyssna's MCP supports semantic search using embeddings. Just describe the topic and it'll find the research that matches.

Here it's found some relevant studies and some common complaints around the shipping cost and the clarity at checkout. These are real quotes from real Lyssna studies, not a generic summary of what checkout problems usually look like. Now you can dig deeper. You can segregate by participant demographics, or you can compare the results with another study from here.

Now you can see here that it's also flagged some relevant feedback across some interviews that we've run. Now, interviews are also part of our MCP, so if you want to understand what came up in a round of interviews without rewatching the recordings yourselves, now you can.

Let's look more at interviews. We can ask, for example, what did participants say about onboarding in our last round of interviews? You'll see it pulling straight from the transcript and the AI summary with quotes attributed to who said them, so that the participants don't get blended into a single impression.

Access to your Lyssna studies follows your Lyssna account permissions, including open and closed spaces, so users only see the research that they're allowed to see. That includes people on our free viewer seats. If you can view the study in Lyssna, then you can also ask about it here. You don't have to be the person who ran it, and now you don't have to ping the researcher and wait on a summary.

One more workflow that I love as a product manager is integration with other tools. If you've got tools like Linear or Notion also connected to your AI assistant, you can combine them in the same conversation. So these insights that we're getting from our interviews can go directly to the rest of the team who needs them when they need them.

Lyssna supplies the evidence, the findings, and the quotes, and your assistant turns that into the ticket or the doc in the tool where the action belongs. No need to copy findings across by hand. Now, one thing I want to be clear on, this isn't the model guessing what users usually think. Every answer you saw from a usability test, a survey, or an interview, that's grounded in actual responses from actual Lyssna studies.

Now, because it's AI, it's still worth double-checking anything that really matters, especially numbers. But our team and many others are already using this day to day and seeing great results.

So that's Lyssna's MCP server, and now generally available. Connect it once and your research is something that you can analyze from Claude, ChatGPT, or any other MCP client you connect yourself. The help docs are linked below for the full setup guide, and we've got much more planned in this space. So if you run into any trouble or you have any feedback, please tell us.

That's what shapes what we build next. Thanks for watching.