Start your free trial
English (United Kingdom) Español Français

How to configure Empower MCP

  • Updated

The Empower Model Context Protocol (MCP) connects your AI assistants directly to your Empower call data, transcripts, and action items. In this guide, you’ll learn how to access your MCP server URL, integrate Empower with Claude or ChatGPT, and configure custom MCP client connections.

Connect and configure Empower Model Context Protocol (MCP)

1. Access the Empower MCP page and copy your server URL

Locate your unique MCP server URL to prepare for AI assistant integration.

  1. Log in to your Empower account.
  2. Navigate to the Empower MCP page.
  3. Click to copy your unique MCP server URL.

2. Connect Empower to Claude

Add Empower as a custom connector in Claude to query your call data directly from chat.

  1. Open Claude and navigate to Customize > Connectors.
  2. Select Add custom connector.
  3. Enter the connector details:
    • Name: Empower
    • MCP Server URL: Paste your copied MCP server URL
    • (Note: Leave the optional OAuth Client ID and Client Secret fields empty)
  4. Click Add, then select Connect.
  5. Sign in to your Ringover account and approve access permissions for Empower.
  6. Start asking questions in Claude (e.g., "What were the key takeaways from my last call?" or "Show me my open action items").

3. Connect Empower to ChatGPT

Set up Empower as a custom developer app in ChatGPT Pro, Business, or Enterprise.

  1. Enable Developer Mode: Go to Settings > Apps > Advanced Settings and turn on Developer Mode (Note: Business, Enterprise, and Edu accounts require workspace admin authorization).
  2. Create Custom App: Navigate to Settings > Apps > Create (or Workspace Settings > Apps > Create), name the app Empower, and paste your MCP server URL.
  3. Scan Tools and Authorize: Select OAuth and Dynamic Client Registration (leave Client ID and Client Secret empty), click Scan Tools, then complete the Ringover sign-in and consent prompt.
  4. Create the App: Click Create after the tool scan finishes. Empower will appear under Enabled Apps with a Dev label.
  5. Publish and Use: Workspace admins can publish the app from Workspace Settings > Apps > Drafts. In chat, @mention or select Empower on any prompt requiring live call data.

4. Configure custom developer integrations (Advanced)

For custom applications, register your public client and handle OAuth authentication with PKCE.

  1. Register Public Client: Send a POST request with your redirect URIs to the advertised registration endpoint. Store the returned emcp_client_... ID (No client secret is issued).
  2. Authorization Request with PKCE: Generate a random state and a 43–128 character verifier, set code_challenge to base64url(SHA-256(verifier)), and direct the user to the authorization URL.
  3. Exchange Code for Tokens: Validate the callback and POST the authorization code with the redirect URI, verifier, and MCP resource URL. Store the returned emcp_access_... and emcp_refresh_... tokens.
  4. Initialize Session & Execute Tools: Initialize the Streamable HTTP session using the access token in the Bearer header, confirm initialization, and run tools/list to call available endpoints.

Check out the dedicated MCP Documentation guide for complete technical references.

FAQ

How do I connect Empower MCP to my AI client?

Paste the MCP server URL shown above and choose OAuth. In Claude, ChatGPT, or another client that supports dynamic registration, leave the optional Client ID and Client Secret fields empty.

Never paste an Empower API key or Ringover token.

Who can connect to Empower MCP?

Each person needs a Ringover account with Empower enabled for their user and team.

Claude, ChatGPT, Cursor, and other remote MCP clients can connect. Some workspaces require an administrator to enable custom connectors first.

Can multiple users connect to the same custom app?

Yes. Authorization is tied to the user’s Ringover identity and permissions, so tokens must never be shared.

A custom app registers once and reuses its client ID, but every user completes their own sign-in and consent.

What data can Empower MCP access?

It can read only the Empower data that the signed-in user may view, such as calls, transcripts, summaries, action items, analytics, meetings, and connected CRM or ATS records.

Empower MCP cannot create, edit, or delete data in Ringover, your CRM, or your ATS.

Why can’t I see or use certain Empower MCP tools?

Available tools and results follow the user’s current Empower plan, role, and Ringover permissions.

After access changes, refresh the connector’s tools or start a new session. A tool that is not listed for that user cannot be called.

What should I do if the MCP connection fails?

First, confirm that you copied the exact regional URL shown above, including /mcp, and selected OAuth.

Access tokens should refresh automatically. If retrying still fails, disconnect and reconnect because the Ringover session or refresh token may no longer be valid.

How is my connection secured?

The connection uses HTTPS and per-user OAuth. Your underlying Ringover token is never exposed to the AI client, and credentials are never accepted in URLs.

Tool results are still handled by your AI provider, so its privacy and retention policies also apply.

How do I disconnect or revoke Empower MCP access?

Remove or disconnect Empower in the AI assistant to clear its stored connection.

To invalidate access immediately on the server, revoke the user’s Ringover session or ask your administrator or Ringover support for help.

🎉 You now know how to connect and configure Empower MCP and can seamlessly query your Empower call insights and action items directly inside Claude, ChatGPT, or custom AI applications!