Plug your AI into the voice of your customer.
The Evermuse MCP gives Claude, ChatGPT, Cursor, and any MCP-compatible agent direct access to your real customer evidence — interviews, sales calls, feedback, quotes, competitor intel, and your live roadmap. Stop describing your customers to your AI. Let it talk to them.
What is the Evermuse MCP?
The Evermuse MCP server plugs your AI assistant straight into your customer knowledge base. Instead of copy-pasting transcripts into ChatGPT or asking Claude to guess, your assistant can now answer real questions about your customers using your actual interviews, sales calls, feedback, and roadmap — and it can pull in quotes, run research, and even take action in the other tools you already use.
Twenty grounded tools. One OAuth-authenticated endpoint. Every AI client that speaks MCP. You stop describing your customers to your AI and start letting it talk to them.
What the Evermuse MCP unlocks
Twenty tools organized into five capabilities your AI can use across every conversation.
Surface customer voice instantly
01 / 05Search everything your customers have ever said in one shot — needs, complaints, quotes, meeting clips, uploaded docs — without remembering where it lives or who said it.
"What are customers saying about onboarding friction?" Your assistant pulls the most relevant quotes, needs, and meeting moments in one call — ranked by meaning, not keywords.
Ground every product decision in real evidence
02 / 05Pull verbatim customer quotes for your next PRD, see live competitor capabilities, check what your team is researching right now, and read full transcripts on demand.
"Find three customer quotes about checkout pain for the pitch deck." Real quotes, attributed to real meetings, dropped into your draft in under a minute.
Move from insight to roadmap
03 / 05Read your live, prioritized roadmap, browse your in-flight shaping work from idea through PRD to ready, and open any feature doc in full — directly from the conversation.
"What's at the top of the roadmap, and which shaping notes are still in the PRD stage?" Answered from the live system, not last week's screenshot.
Bring evidence in, push work out
04 / 05Drop new transcripts, notes, and documents into the right research project from inside any AI conversation — then reach into the tools you've already connected (Linear, Jira, Slack, HubSpot, your own MCP servers) to create tickets, send updates, and take real action.
An AE wraps a Gong call and says "save this transcript to the Q2 discovery project, then open a Linear ticket for the pricing pushback." Both happen, in one turn.
Hand off the heavy lifting
05 / 05For deep questions that would drown a normal chat, spin up a focused research worker that returns a clean, cited answer — without polluting your main conversation with raw transcripts.
"Read every enterprise call last month and tell me the top three reasons we lost." The subagent does the reading; you get back a structured set of findings with confidence levels and underlying evidence.
Install the MCP in your agent
One OAuth 2.1 endpoint. Pick your client and follow the steps.
https://api.evermuse.com/api/mcpClaude Desktop
Connect via the built-in MCP connector.
- 1
Open Claude Desktop → Settings → Connectors
- 2
Click "Add custom connector"
- 3
Paste the Evermuse MCP URL
https://api.evermuse.com/api/mcp - 4
Authenticate via OAuth using your Evermuse account
- 5
Start a new chat and ask: "What are customers saying about onboarding?"
Claude Code
One-line install via the CLI.
- 1
In your terminal, run:
claude mcp add --transport http evermuse https://api.evermuse.com/api/mcp
Cursor
Add via Settings → MCP, or drop into your MCP config.
- 1
Open Cursor → Settings → MCP → Add new MCP server
- 2
Name it "evermuse" and set the URL to:
https://api.evermuse.com/api/mcp - 3
Choose OAuth as the auth method and sign in
{
"mcpServers": {
"evermuse": {
"url": "https://api.evermuse.com/api/mcp"
}
}
}ChatGPT
Add as a custom connector in ChatGPT (Pro/Business workspaces).
- 1
Open ChatGPT → Settings → Connectors → Add custom connector
- 2
Paste the Evermuse MCP URL
https://api.evermuse.com/api/mcp - 3
Authenticate via OAuth using your Evermuse account
Codex / other MCP clients
Any OAuth 2.1 MCP-compatible client works.
- 1
In your client's MCP config, add the Evermuse server with transport "http" and the OAuth URL below
- 2
Authenticate when prompted — Evermuse uses standard OAuth 2.1
{
"mcpServers": {
"evermuse": {
"transport": "http",
"url": "https://api.evermuse.com/api/mcp"
}
}
}Why this beats "just ask ChatGPT"
A general AI doesn't know what your customers said in last Tuesday's call. The Evermuse MCP does.
- —Confident answers built on whatever your AI happens to "remember"
- —Copy-pasting transcripts into chats and hoping they fit in the context window
- —Stale screenshots of the roadmap referenced in conversations
- —No verifiable source behind any quote, claim, or recommendation
- —Disconnected tools — your AI can't actually do anything in Linear, Slack, or Jira
- Answers grounded in your actual interviews, calls, and feedback
- One unified semantic search across every customer signal you've captured
- Live roadmap and shaping notes — always the current version
- Every quote and claim citeable back to a real meeting or document
- Two-way action: drop in new evidence, create tickets, send updates — without leaving the chat