IceCubes API and MCP Server: Build Meeting Intelligence into Any Workflow

Meeting data is trapped. It sits inside meeting platforms, note-taking apps, and CRM records - accessible through UIs designed for humans, not systems. If you want to build a custom deal health dashboard that incorporates meeting insights, or trigger a workflow when specific topics come up in customer calls, or feed meeting context to an AI assistant, you need programmatic access.

IceCubes provides this through two interfaces: a REST API for traditional integrations and an MCP (Model Context Protocol) server for AI tool integrations. This post covers both, with practical use cases for each.

The REST API

The IceCubes API gives you access to meeting transcripts, AI-generated insights, and user data through standard REST endpoints. Authentication uses API keys tied to your IceCubes account.

Core Endpoints

Meetings

• GET /api/v1/meetings - List meetings with filtering and pagination
• GET /api/v1/meetings/:id - Get a specific meeting with full transcript
• GET /api/v1/meetings/:id/transcript - Get just the transcript
• GET /api/v1/meetings/:id/insights - Get AI-generated insights (summary, action items, MEDDIC, Smart Tags)

Search

• GET /api/v1/search - Search across meeting transcripts by keyword, date range, participant, or tag

Insights

• GET /api/v1/insights/action-items - All action items across meetings, filterable by assignee and status
• GET /api/v1/insights/smart-tags - Smart Tag results across meetings

Response Format

All endpoints return JSON. A meeting response includes:

{
  "id": "meeting_abc123",
  "title": "Q1 Pipeline Review - Acme Corp",
  "date": "2026-02-20T14:00:00Z",
  "duration_minutes": 42,
  "platform": "google-meet",
  "participants": ["Sarah Chen", "John Martinez", "Lisa Park"],
  "summary": "Discussed Q1 pipeline targets...",
  "action_items": [
    {
      "text": "Send updated pricing proposal",
      "assignee": "John Martinez",
      "due_date": "2026-02-22",
      "status": "pending"
    }
  ],
  "smart_tags": [
    {
      "tag": "Budget Timing",
      "matches": [
        {
          "speaker": "Lisa Park",
          "text": "Our fiscal year ends in March, so we need to get this approved before then",
          "timestamp": "00:18:34"
        }
      ]
    }
  ]
}

Pagination and Filtering

List endpoints support standard query parameters:

Parameter	Description	Example
limit	Results per page (max 100)	?limit=25
offset	Skip N results	?offset=50
date_from	Filter by start date	?date_from=2026-01-01
date_to	Filter by end date	?date_to=2026-02-28
participant	Filter by participant name	?participant=Sarah+Chen
platform	Filter by meeting platform	?platform=zoom
smart_tag	Filter by Smart Tag	?smart_tag=Budget+Timing

The MCP Server

MCP (Model Context Protocol) is an open standard for connecting AI tools to external data sources. IceCubes provides an MCP server that lets AI assistants like Claude, ChatGPT, and custom LLM applications access your meeting data directly.

Why MCP Matters

Without MCP, getting meeting context into an AI conversation requires manual copy-pasting. "Here's the transcript from my meeting, what should I do next?" With MCP, the AI tool can query your meetings directly:

• "What did the client say about their budget in our last three calls?"
• "Summarize all action items assigned to me this week"
• "What competitors have been mentioned in our sales calls this month?"

The AI tool calls the IceCubes MCP server, retrieves the relevant data, and responds with context-aware answers - no copy-pasting required.

MCP Tools Available

The IceCubes MCP server exposes the following tools to connected AI clients:

• search_meetings - Find meetings by keyword, date, participant, or tag
• get_meeting - Retrieve full meeting details including transcript
• get_insights - Get AI-generated insights for a specific meeting
• get_action_items - List action items with filters
• get_smart_tags - Query Smart Tag results across meetings

Setting Up the MCP Server

For Claude Desktop, add the IceCubes MCP server to your configuration:

{
  "mcpServers": {
    "icecubes": {
      "command": "npx",
      "args": ["icecubes-mcp-server"],
      "env": {
        "ICECUBES_API_KEY": "your-api-key"
      }
    }
  }
}

Once configured, you can ask Claude questions about your meetings naturally, and it will use the MCP tools to retrieve relevant data.

Use Case: Custom Deal Health Dashboard

Problem: Your CRM shows deal stages, but managers want to see meeting-derived signals - how many discovery questions were asked, which MEDDIC fields are qualified, what objections remain unresolved.

Solution with the API:

1. Poll GET /api/v1/meetings daily for new meetings tagged with CRM opportunity IDs
2. Extract MEDDIC scores, objection counts, and engagement metrics from /insights
3. Aggregate into a custom dashboard (Grafana, Retool, or your own frontend)
4. Display alongside CRM pipeline data for a complete deal health view

Key metrics to extract:

Metric	Source	Signal
MEDDIC completion	insights.meddic	Qualification depth
Unresolved objections	insights.objections	Deal risk
Competitor mentions	insights.competitors	Competitive pressure
Action item completion	insights.action_items	Follow-through
Meeting frequency	meetings list	Engagement level
Stakeholder count	participants	Multi-threading

Use Case: Automated Follow-Up Pipeline

Problem: After every customer call, reps should send a follow-up email summarizing key points and next steps. This takes 15-20 minutes per call and often doesn't happen.

Solution with the API:

1. Set up a webhook listener for the meeting.completed event
2. When a meeting ends, fetch the summary and action items via the API
3. Use a template engine to format a follow-up email
4. Route to the rep for review and send (or auto-send for routine meetings)

This reduces follow-up time from 15 minutes to 30 seconds (reviewing the draft) and ensures it happens for every call, not just the ones where the rep remembers.

Use Case: AI-Powered Prep for Upcoming Meetings

Problem: Before a call with a prospect or customer, reps need to review past interactions. This usually means scrolling through CRM notes and trying to remember what happened.

Solution with MCP:

Before a meeting with Acme Corp, ask your AI assistant:

• "Summarize our last 3 meetings with Acme Corp"
• "What action items are still open from our previous calls?"
• "What concerns has Lisa Park raised across our conversations?"

The MCP server retrieves the relevant meetings and the AI synthesizes a briefing. This takes 60 seconds instead of 15 minutes of manual research.

Use Case: Weekly Intelligence Reports

Problem: Sales leadership wants a weekly report on competitive mentions, common objections, and deal progression signals across the entire team.

Solution with the API:

1. Weekly cron job queries GET /api/v1/insights/smart-tags for the past 7 days
2. Aggregate competitor mentions, objection types, and buying signals
3. Generate a formatted report (Markdown, PDF, or Slack message)
4. Distribute to leadership automatically

This replaces the manual process of asking each manager to compile their team's competitive intelligence.

Use Case: Internal Knowledge Base Enrichment

Problem: Product decisions, customer feedback, and strategic context get discussed in meetings but never make it into your wiki or knowledge base.

Solution with the API + Zapier:

1. When a meeting's Smart Tags include "Feature Request" or "Product Decision," trigger a Zapier workflow
2. Extract the relevant transcript sections via the API
3. Create a draft page in Notion, Confluence, or your wiki
4. Tag the relevant product manager for review

Meeting knowledge stops being ephemeral and becomes part of your institutional memory.

Rate Limits and Best Practices

• API rate limits are 100 requests per minute per API key
• Use pagination for large result sets rather than requesting everything at once
• Cache meeting data locally if you're building a dashboard that refreshes frequently
• Use webhook events for real-time workflows rather than polling

Getting Started

The IceCubes API and MCP server are available on all plans. Generate your API key in Settings > Integrations > API. Full API documentation is available in the IceCubes dashboard.

If you're not yet using IceCubes, start with 50 free AI credits - no credit card required. Install the browser extension to start building your meeting data foundation.

Add to Chrome | Add to Edge