Make.com AI Agent Tutorial – Build Your First Agent in 20 Minutes

Overview

This guide follows the 20‑minute Make.com video to build a simple AI agent that can answer inventory questions and place orders. The agent uses an LLM (OpenAI or Grok), two tool scenarios (list inventory and order stock), and a Slack trigger to interact with users.

Prerequisites

• A Make.com account with a paid plan (free plan limits operations).
• An API key for an LLM provider (OpenAI or Grok).
• A Slack workspace with a bot token (or other input channel).
• A Google Sheet or Airtable with inventory data.

Step‑by‑Step Instructions

1. Create the AI Agent

1. In Make, navigate to AI Agents → Create agent.
2. Name: Inventory Assistant.
3. Description: You help manage store inventory. You can check stock levels or place restock orders using the tools provided. Only respond based on tool outputs and available data.
4. Connection: Choose your LLM provider, paste the API key, and save.

2. Build Tool Scenarios

2.1 List Inventory

1. Create a new scenario named List Inventory.
2. Add a Google Sheets module (or Airtable) to read the inventory sheet.
3. Map the optional Product Name input (if any).
4. End with a Return module: output Stock Details (text).
5. Set trigger to On demand and activate.

2.2 Order Stock

1. Create a scenario named Order Stock.
2. Add modules to create a purchase order (e.g., Google Sheets row creation or email to supplier).
3. Inputs: Product Name, Quantity.
4. End with a Return module: output Confirmation Message.
5. Set trigger to On demand and activate.

3. Attach Tools to the Agent

1. In the agent settings, go to Tools → Add.
2. Select the two scenarios created above.
3. Ensure each has a clear description (e.g., Returns stock levels for products).

4. Connect the Agent to Slack

1. Create a new scenario.
2. Trigger: Slack > Watch messages.
   • Choose channel or DMs.
   • Add a filter to exclude bot_message subtypes.
3. Action: Make AI Agents > Run an agent.
   • Select Inventory Assistant.
   • Map Message field to Slack message text.
4. Reply: Slack > Create a message.
   • Map Response from the agent to the message content.
   • Use the original message’s Thread Timestamp to reply in thread.
5. Activate the scenario.

5. Test the Agent

1. Send a Slack message like Do we have blue hoodies in stock?.
2. Verify the agent calls List Inventory and posts the result.
3. Send Order 10 more bags of coffee.
4. Verify the agent calls Order Stock and posts the confirmation.

Tips & Gotchas

• Tool Descriptions: The agent relies on tool names/descriptions to decide which scenario to run. Keep them concise and user‑friendly.
• Return Module: Every tool scenario must end with a Return module; otherwise the agent cannot use the output.
• Memory: Make agents do not retain state between runs. If you need context, pass it explicitly in the prompt.
• Slack Bot Filtering: Without filtering out bot_message, the agent will trigger on its own replies and create loops.
• LLM Costs: Each agent run counts as an LLM operation. Monitor usage to avoid unexpected charges.
• Testing: Use the scenario run history to debug which tool was selected and what inputs/outputs were used.

This markdown follows the exact steps shown in the Make.com video, ensuring a reproducible setup for a functional inventory AI agent.