With Lyssna's MCP server, you can connect Claude, ChatGPT, and other AI assistants directly to your Lyssna research data and ask questions about your studies to get answers grounded in what participants actually said.
The integration is read-only: it can answer questions grounded in your existing study data, but it can't create studies or write or edit anything back into Lyssna. Tool accuracy is still being tuned as we continue to improve the feature.
What you can do with it
Once connected, you can use your AI assistant to explore and analyze your Lyssna research data in a number of ways:
Ask a question about a single study, or across every study you have access to, and get an answer pulled from real response data, not a guess.
Compare results across studies (for example, "compare Test A and Test B and tell me why one outperformed the other") instead of reopening each one to find what you need.
Check whether something's already been tested before commissioning new research.
Bring findings straight into whatever you're already working on: a PRD, a design rationale, a stakeholder update.
Combine Lyssna's research data with other MCP-connected tools you use, so you can turn a finding into action in the tool where that action belongs.
What data it can access
The MCP server has read-only access to the following data from your unmoderated tests and surveys, and moderated interviews:
Basic study details
Study structure (sections, tasks, questions, assets)
Recruitment info
Responses, answers and follow-ups
Screener answers
Recordings
Access always follows your existing Lyssna permissions, so you'll only see the tests and studies you already have access to in Lyssna itself.
Setting up the integration
Go to your profile settings page and find the MCP integration options under Integrations.
Click Connect next to the AI assistant you use
Follow the below instructions for the AI assistant you're connecting.
Claude
Claude
Follow this link to add a custom connector, adding these details:
Name: Lyssna
Remote MCP server URL: https://mcp.lyssna.com/mcp
Leave Advanced settings blank
Press "Add" to save and continue
Press connect on your new connector
Press "Allow access" to connect Claude to your Lyssna account
Claude Team or Enterprise
On a Claude Team or Enterprise plan, an Owner or Primary Owner adds the connector once for the organisation, then each team member connects to it. Steps match Anthropic's Team and Enterprise custom connector flow.
Owner or Primary Owner:
Navigate to Organization settings → Connectors.
Click Add.
Hover over Custom, then select Web.
Add the remote MCP server URL:
https://mcp.lyssna.com/mcpLeave Advanced settings blank (no OAuth credentials needed).
Click Add to finish.
Team members (once the Owner has added it):
Navigate to Customize → Connectors.
Find the Lyssna connector in the list (it will have a "Custom" label).
Click Connect to authenticate.
ChatGPT
ChatGPT
Log in to ChatGPT, navigate to the Lyssna plugin, and press "Install plugin"
Press "Allow access" to connect ChatGPT to your Lyssna account
If your workspace uses custom MCP connectors / developer mode instead of the plugin listing, see OpenAI's guide: Developer mode and MCP apps in ChatGPT. Use server URLhttps://mcp.lyssna.com/mcpand OAuth when prompted.
Cursor
Cursor
Open Cursor Settings → Tools & MCP.
Click New MCP server.
Paste this into the
mcp.jsonfile that opens:
{
"mcpServers": {
"lyssna": {
"type": "http",
"url": "https://mcp.lyssna.com/mcp",
"auth": {
"CLIENT_ID": "lyssna-cursor"
}
}
}
}Back on the Tools & MCP page, the Lyssna server appears with an Authenticate button. Click it and follow the prompts.
Gemini
Gemini
Gemini (personal accounts)
Go to gemini.google.com/apps and add a custom app.
Enter the server URL:
https://mcp.lyssna.com/mcpApprove access when Lyssna's consent screen appears.
Custom apps are a consumer Gemini feature. They aren't available on a Google Workspace business or education account, and they need a Google AI Pro or Ultra subscription.
Gemini registers connectors dynamically, so the name on the Lyssna consent screen is set by whoever started the connection. Only approve a connection you started yourself, from the tab you started it in.
Gemini Enterprise
Set up by your Google Cloud admin, once for your organisation. Gemini Enterprise is a separate product from the Gemini chat at gemini.google.com — Workspace Gemini alone isn't enough.
Google's guides:
Set up your custom MCP server data store. Use the Lyssna values below wherever that guide asks for your MCP / OAuth endpoints.
Turn on the tools (Enable actions)
Give people access. Each user needs:
The Gemini Enterprise User IAM role (
roles/discoveryengine.agentspaceUser) on the project or app — see IAM roles and permissions and Configure access controls for apps.A licence seat on your Gemini Enterprise subscription — assign manually or automatically under Gemini Enterprise → Manage users, as described in Get subscriptions and assign licenses for Gemini Enterprise (and licence FAQs if access still fails after assignment).
The first time they use the Lyssna connector in chat they'll sign in to Lyssna individually, and every tool call runs under their own Lyssna permissions.
Details you'll need
MCP server URL:
https://mcp.lyssna.com/mcpAuthorization URL:
https://app.lyssna.com/oauth/authorizeToken URL:
https://app.lyssna.com/oauth/tokenEnable PKCE support: on
Microsoft 365 Copilot
Microsoft 365 Copilot
This is set up by your Microsoft 365 admin, once for your whole tenant. Individual users can't add it themselves.
Microsoft's end-to-end guide for custom MCP connectors: Set up custom federated connectors. Related: Register MCP servers as agent connectors and OAuth authentication for plugins.
Before you start: everyone who will use Lyssna in Copilot needs a Microsoft 365 Copilot licence. Without it, Copilot may still let them turn the connector on and sign in to Lyssna, then quietly not use it. Assign licences in the Microsoft 365 admin centre — see Set up Microsoft 365 Copilot and Assign licenses to users.
Details you'll need
MCP server URL:
https://mcp.lyssna.com/mcpAuthorization endpoint:
https://app.lyssna.com/oauth/authorizeToken endpoint:
https://app.lyssna.com/oauth/tokenRefresh endpoint:
https://app.lyssna.com/oauth/tokenScope: leave blank
PKCE: must be enabled
Client ID:
copilot-6ba42c3f2e7ac0a85ccf2c6330cac268Client Secret:
oxH5J9klGOXgN5s057IAVHDbCSEwHSmQSYYUFua-oMYiOYUXasNEZ3WUyb45_Hp6Redirect URI (if your provider asks):
https://teams.microsoft.com/api/platform/v1.0/oAuthRedirect(per Microsoft's OAuth redirect requirement)
For end users:
Go to Settings → Sources
turn on the Lyssna connector, and sign in to Lyssna when prompted.
Other MCP Clients
Other MCP Clients
Lyssna is published in the official MCP registry. Any client that supports remote MCP servers can connect.
Server URL:
https://mcp.lyssna.com/mcpTransport: Streamable HTTP
Authentication: OAuth 2.1 authorisation code with PKCE (required)
Most clients discover everything else from https://app.lyssna.com/.well-known/oauth-authorization-server.
Add the server URL in your client's MCP settings, then complete the Lyssna sign-in prompt.
Asking questions
Once you're connected, just ask your AI assistant a question about your collected data. A few examples:
"What have participants said about our checkout flow across the last three studies?"
"Have we already researched anything about our onboarding flow? What did we learn?"
"Summarize the key findings from our latest usability test on our homepage."
"Compare Test A and Test B and tell me why one outperformed the other."
If the first answer isn't specific enough, ask a follow-up, narrow it down, or dig into one particular study.
Every answer is grounded in your actual study data, but we recommend double-checking anything critical as AI can make mistakes.
Disconnecting
To disconnect an assistant from your Lyssna account, visit your profile settings, open the dropdown next to the connected assistant, and select Disconnect.
Workspace admins can also turn the whole MCP integration off for the account at any time from Settings, without needing us to change anything.
What happens when I disconnect?
When you disconnect an assistant, here’s what to expect:
Lyssna will no longer share your research data with that assistant.
Any conversation history already in your AI assistant isn't affected, since we don't control what happens on that side.
You can reconnect at any time from your profile settings.
Common questions
Is the MCP integration available on all plans?
The MCP integration is available on all paid plans, including legacy plans.
Can MCP create or edit anything in my Lyssna account?
No — the integration is read-only. It can answer questions based on your existing study data, but it can't create studies, add responses, or edit anything in your account.
Which AI assistants are supported?
Claude, ChatGPT, Microsoft 365 Copilot, Gemini, and Cursor, plus any other client that supports remote MCP servers. See the setup sections above.
Can my whole team use the integration once it's connected?
Connections are per user. Each person connects (or signs in) with their own Lyssna account and only sees studies they already have access to. Org-level setup for Copilot and Gemini Enterprise is done once by an admin; individuals still complete Lyssna sign-in.
Can admins disable MCP for the whole workspace?
Workspace admins can turn the MCP integration off for the entire account at any time from Settings → Feature access.
I'm on a Claude Team or Enterprise plan — can I set up the custom connector myself?
Not directly — there's a two-step process for Team and Enterprise plans.
Preliminary steps for the Owner or Primary Owner:
Navigate to Organization settings > Connectors.
Click the "Add" button.
Hover over "Custom," then select "Web."
Add the remote MCP server URL: https://mcp.lyssna.com/mcp
Leave Advanced settings blank (no OAuth credentials needed).
Click "Add" to finish.
Steps for team members (once the Owner has added the connector):
Navigate to Customize > Connectors.
Find the Lyssna connector in the list (it will have a "Custom" label).
Click "Connect" to authenticate and start using it.
Running into an issue, or have feedback on what you'd like MCP to do next? Reach out to our support team.





