Use Intellicoach from Claude and ChatGPT
Intellicoach is an MCP server. Connect it once and your AI assistant can read your DM pipeline, pull transcripts, explain why the AI said what it said, and change your scripts and settings when you ask. Your assistant never writes or sends a message itself. The one action that can lead to a message is turning your AI setter on for a conversation, which lets the AI reply to that prospect, and it only happens when you ask.
Claude and ChatGPT both connect with this URL:
https://app.intellicoach.ai/api/mcp1. Connect in Claude
- In Claude, open Customize, choose Connectors, then select Add and Add custom connector.
- Name it Intellicoach, paste the URL above, and add the connector.
- Click Connect. You are sent to Intellicoach to sign in with your account email and the code we email you.
- Review the permissions, then approve. Read and write permissions are listed separately.
- Start a chat, enable the Intellicoach connector, and ask a question.
Works on Claude.ai, Claude Desktop, Claude mobile, Claude Code and Cowork.
2. Connect in ChatGPT
- Use ChatGPT in a browser, open Plugins, and choose New Plugin from the plus menu. If that option is missing, turn on developer mode under Settings, Security and Login.
- Name the plugin Intellicoach, paste the URL above, and create it.
- Select Sign in with Intellicoach, sign in, and approve the permissions.
- Mention Intellicoach in a chat or pick it from the plugin list.
Works in ChatGPT and Codex.
3. Other MCP clients
Any client that supports remote MCP over streamable HTTP with OAuth 2.1 can connect. Point it at the URL above. Authentication is the authorization-code flow with PKCE (S256); the server supports Client ID Metadata Documents and dynamic client registration, and accepts loopback redirects for desktop clients. Discovery metadata is at /.well-known/oauth-protected-resource and /.well-known/oauth-authorization-server on app.intellicoach.ai.
Claude Code: claude mcp add --transport http intellicoach https://app.intellicoach.ai/api/mcp
What your assistant can do
Read
- Pipeline snapshot, funnel movement and analytics for any period
- Active conversations, stage lists, and who needs a response
- Full DM transcripts, with who sent each message and what stage the AI was in
- Follow-up queue, scheduled reminders and never-replied contacts
- Your scripts, Framework SOP, objection playbook, persona and response settings
- Keywords, resources, testimonials, tag rules, memories and voice clips
- Bookings with call times, provider status and recorded outcomes
- Diagnostics for one conversation: decision trace, delivery events, known facts, what changed since
Change, when you ask
- Edit scripts, objection playbook, persona and response settings (every change is logged and undoable)
- Move leads between stages, set categories and tags, write notes
- Schedule, complete or dismiss follow-ups and record call outcomes (marking a call booked or closed also records the conversion with your connected Meta pixel and CRM)
- Turn the AI on or off for a specific conversation (turning it on lets your AI setter reply to that prospect right away)
- Create, edit, enable, disable or delete keywords, resources, testimonials and tag rules
- Protect a contact from AI outreach
Never
- Write or send a message to a prospect itself
- Replace a whole script or document in one step (long text changes are passage edits anchored on the current wording)
- Bulk-edit more than 100 conversations, or edit by filter instead of an explicit list
- Touch billing, team members, channel connections or credentials
Things to try
- Give me my DM briefing for the last 7 days: what moved, what is stuck, and who needs a reply.
- Show me the conversations in Closing that have not booked yet, and summarize each in one line.
- Pull up my conversation with @sam_transform and tell me why the AI asked about budget twice.
- What does my qualification script ask right now? Suggest one tighter version of question 3 and preview the change before applying it.
- Schedule a follow-up with Jamie Smith for next Tuesday with the note: check in after her assessment.
- Which keyword automations are disabled, and how many conversations did each one start this month?
Permissions and control
- You approve each permission on the Intellicoach consent screen. Write permissions are shown separately and are only granted when the assistant requests them.
- Writes happen only when you explicitly ask. Settings edits are previewed before they apply, recorded in your settings history, and can be undone.
- Every connection appears in Settings, then Connected apps inside Intellicoach, where you can revoke it at any time.
- Access tokens expire after one hour and refresh automatically; a connection that goes unused for 90 days expires on its own.
- Every call is audit logged with the tool name and argument names, never the values.
Your data
When you ask your assistant a question, the information it needs to answer (for example a transcript, a pipeline list or your script) is sent from Intellicoach to that assistant's provider, such as Anthropic or OpenAI, and is then handled under that provider's terms. Intellicoach only returns what the specific request needs, and never returns access tokens, internal identifiers you did not ask for, or another coach's data. See our Privacy Policy and Terms of Service.
Support
Questions or problems connecting? Visit Support or email support@intellicoach.ai. Include which assistant you are using and the name of the tool or prompt that did not work.