Airspeed MCP Server: Revenue Data in Any AI Assistant

Model Context Protocol

Give reps the full deal context in their AI tools.

The Airspeed MCP server brings your calls, playbooks, and agent runs into the tools reps already use. Find the call, score the deal, and draft the follow-up in natural language, grounded in what the buyer actually said.

Connect in 2 minutes → See the use cases
Works with ClaudeCursorChatGPTAny MCP client
For reps. For RevOps.

Reps move faster. Ops keeps control.

Reps ask in plain language. The answers come from the calls and playbooks your Ops team already governs in Airspeed.

Ops controls the keys

The server authenticates with an Airspeed API key from your settings. A key carries org-wide read access to calls, playbooks, and agent runs, so your team decides who holds one.

One playbook, every rep

Define scoring criteria once in Airspeed. Every assistant answers from the same playbook definition, with full version history behind it.

Answers you can trace

Every summary cites the call, snippet, or agent run it came from, so any claim is one step from the conversation that proves it.

Use case 01 · Call prep

Walk in with the whole history.

Ask for every prior conversation with the account, the open objections, and the next step you agreed. The assistant pulls it from real calls in Airspeed, not from memory.

list_callsget_call_infoget_call_snippets
Your AI assistant

"Pull my last call with ThetaRay and list every objection and the next step we agreed."

Answered from your Airspeed calls, with the source cited.

Use case 02 · Deal scoring

Score any deal against your playbook.

Ask what is missing on your enterprise criteria. The assistant reads the playbook definition your Ops team maintains and checks the deal’s conversations against every criterion.

list_playbooksget_playbookget_call_info
Your AI assistant

"Score the Fingercheck deal against our enterprise playbook and tell me what’s missing."

Answered from your Airspeed calls, with the source cited.

Use case 03 · Follow-up

Draft the follow-up from what was said.

Turn today’s call snippets into a recap and a follow-up email in the buyer’s own words, and pull the structured output of any agent run into the same thread.

get_call_snippetsget_call_mediaget_agent_run
Your AI assistant

"Draft a follow-up email from the snippets in today’s Runway call."

Answered from your Airspeed calls, with the source cited.

See how reps run their whole day in Airspeed →

Setup

Connect the server.

Add Airspeed to your MCP client's config, authenticate once, and the tools appear.

  1. 1

    Add the server

    Paste the config into your MCP client (Claude, Cursor, or any MCP-compatible app).

  2. 2

    Add your API key

    Paste an API key from your Airspeed settings. Keys carry org-wide read access, so treat them like credentials.

  3. 3

    Start asking

    The tools load automatically. Ask about any call, deal, or playbook.

mcp config (e.g. Claude Desktop) 
{
  "mcpServers": {
    "airspeed": {
      "url": "https://api.glyphic.ai/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}
Available tools

What the assistant can do.

Eleven tools across calls, playbooks, and agent runs, the same objects behind the API.

Calls

  • list_calls

    Find conversations by date, owner, deal, or tag.

  • get_call_info

    Full detail for a call: participants, metadata, extracted fields.

  • get_call_snippets

    Key moments and quotes from the conversation.

  • get_call_media

    Recording and transcript for a call.

  • list_call_tags

    The tag taxonomy used to classify conversations.

Playbooks

  • list_playbooks

    All playbooks available to score deals against.

  • get_playbook

    A playbook definition and its criteria.

  • list_playbook_versions

    Version history for a playbook.

  • get_playbook_version

    A specific historical version of a playbook.

Agent runs

  • list_agent_runs

    Recent agent runs and their status.

  • get_agent_run

    The structured output of a specific agent run.

Keep going

Make the whole team run this way.

For RevOps How to build AI agents for RevOps Standardize the workflows behind the prompts: the data foundation, governance, and where to start. Read the guide → For reps How to build AI agents for account executives Practical workflows that prep calls, keep the CRM clean, and draft outreach that sounds like you. Read the guide →

Frequently asked questions

What does the Airspeed MCP server do?

It gives reps full deal context inside Claude, Cursor, ChatGPT, or any MCP-compatible app. The assistant pulls prior conversations, open objections, and agreed next steps from real calls in Airspeed rather than from memory.

How do you connect the Airspeed MCP server?

Paste the config into your MCP client, then paste an API key from your Airspeed settings. The tools load automatically, and you can ask about any call, deal, or playbook from there.

How is access controlled?

Through the API key. A key carries org-wide read access to calls, playbooks, and agent runs, so treat it like a credential and decide deliberately who holds one.

Can I trust what the assistant tells me?

Every summary cites the call, snippet, or agent run it came from, so any claim is one step from the conversation that proves it. Scoring criteria come from the playbook definition your Ops team maintains, with full version history behind it.

What can the assistant actually do?

Find conversations by date, owner, deal, or tag. Pull full call detail including participants, metadata, and extracted fields. Surface key moments and quotes. Read playbooks and specific historical versions. And fetch the structured output of any agent run.

Model Context Protocol

Bring Airspeed to where your reps already work.

Connect the MCP in minutes, or see it running on your own calls first.

Book a demo → Connect the MCP Talk to our AI rep

Explore Airspeed

Most AI sales tools tell you what happened on the call. Airspeed does the work after it: your pipeline worked, your CRM updated, your reps coached, your forecast called.