Start with ready-made AI agents with instructions on how to manage them on the marketplace. Browse the library

Connecting an External Service via API or MCP

Continuing the case study: connect an external service — an Airtable client base — to a working agent, and compare API and MCP integration in practice.

ASCN Team
11 June 2026

Andres from Mexico City has already set up his morning briefing (Lessons 4.1–4.6): the agent collects emails, checks the calendar, and sends a daily summary every morning. But his workflow goes further — he manages his entire client database in Airtable. Now Andres wants the agent to be able to read that database and update it.

Built-in integrations cover the most common tools. But every business has its own stack: Airtable, HubSpot, Trello, Shopify, Pipedrive, or any other service. For these cases, there are two paths: connecting via MCP or via API.

Path A — via MCP (easier)

An MCP server is a ready-made plugin that the agent knows how to read on its own. You simply connect it — and the agent understands what to do.

Steps:

  1. Find an MCP server for your service on Smithery.ai or GitHub
  2. Copy the server configuration — it's a JSON file with a URL and settings
  3. In ASCN: Brain → integrations section → paste the config
  4. Write to the agent: "You now have access to [service]. Try [test action]"
  5. The agent will discover the tool's capabilities on its own and confirm the connection

Path B — via API (for any service)

The full version of this lesson is available only with the Personal / VIP (365 day's) / Gold (180 day's) / Start / Business plan. Click the button and subscribe to access all lessons.
Choose a plan
By continuing to use our site, you agree to the use of cookies.