Unit Introduction
You’ve reached unit 2 of the MCP course.
In the previous unit, you’ve learned the theory of MCP. Now it’s time to see how this works in Make, starting with the Make MCP client.
You will learn: what you can do with the Make MCP client
when to use it and what benefits it offers
how to use the Make MCP client
Make MCP client
Make provides an MCP client.
The Make MCP Client lets you use tools from MCP servers, so you can connect to external services and tools without writing code or setting up custom integrations.
This means that Make has built its own MCP client that allows you to connect to any third-party MCP server and use their tools directly in your AI agents and scenarios. You can connect to Salesforce’s MCP server so your AI agent can look up customer information and update records in real-time. You can also use it in your Make scenario to automatically sync leads and track deal progress.
The Make MCP client makes it easy to use tools and services that aren’t part of Make’s standard apps.
✅ Here’s why you should use it:
Server selection Select from Make’s list of verified MCP servers or enter a custom server URL.
🚀 Quick setup Connect to tools just by creating a connection, no manual API configuration is required.
🔐 Secure connections Use OAuth to connect to MCP servers without entering passwords directly.
⚡ Tool visibility Once connected, all available tools appear with information about their required and optional fields. In Make, you can use the Make MCP client in two ways:
With your AI agents
Within your Make scenarios
Let’s look at each one.
MCP tools for AI agents
You can connect your AI agent to MCP servers and give it access to additional tools.
When adding tools to your AI agents, you have three options:
1 Make scenarios
2 Make modules
3 MCP
MCP is an easy and straightforward way to access tools from external services.
When to use it
When should you use MCP tools versus modules or scenarios in your AI agents?
It depends on how complex your task is.
MCP tools 👆 Simple action The MCP tool exists, and you want to perform simple and straightforward action.
Example: create this task in Asana if it’s not already present.
Scenario tool or module tool
🤝Combine AI decisions with deterministic workflows Scenarios let you mix the best of both approaches: the AI Agent uses non-deterministic decision-making to choose when to trigger an action based on context, while the scenario uses deterministic execution to ensure the steps happen in a fixed, reliable order.
Example: The AI Agent decides a refund is needed, then triggers a scenario that always follows the exact sequence: verify account → process refund → update CRM → send email.
🔒 Privacy and confidential information Use scenarios when you need to control what data the AI can access. Scenarios let you share only the necessary information, hiding confidential projects, workspaces, or personal data. In contrast, MCP tools make all available data accessible to the AI with no filtering.
Example: For medical appointments send only patient IDs to the AI, not the full medical records. ⚠️ Critical actions requiring extra control Use scenarios when working with important data because they provide safety features that help prevent accidental deletion or modification. With scenarios you can add validation checks that stop the system from executing dangerous operations.
Example: for a GDPR account deletion the scenario checks if account is safe to delete before proceeding.
🔀 Complex data mappings When you need to create records by matching IDs across different systems, scenarios help with handling the data.
Example: Moving leads from Hubspot to Salesforce when the same company has different IDs in each system.
General rule
If your task is straightforward, use MCP tools.
If it’s complicated, handles sensitive data, or requires specific steps in a specific order,ppp,use module tools or build a scenario instead.
When you use MCP tools in your AI Agent, they come directly from the MCP server with their exact descriptions and input/output structure. You’ll work with them exactly as the server provides them, which means you don’t have the option to customize how they’re described or modify their parameters and responses.
MCP tools
You cannot change or customize tool descriptions, inputs, or outputs. The AI Agent works with the exact descriptions, parameters, and data structure provided by the MCP server.
Make scenarios
You have full control to write the tool descriptions, select which input fields to show, and filter outputs to return only relevant and secure data. This lets you personalize how the AI Agent understands and uses the tools while maintaining control over sensitive information.
Real-life examples
Let’s have a look at two real-life examples.
💰 Customer invoice management AI agent 🎯 Objective Build an AI agent that automatically handles customer payments for an online shop. The AI agent will investigate payment issues, identify problems (duplicate charges, failed payments), and create tasks for the team to process refunds.
🏗️ How to do it Connect your AI Agent to MCP servers: Stripe MCP server for payment operations (search customers, list charges, create refunds) Asana MCP server for for task management (create refund tasks and check task status)
🎬 See it in action A customer emails: I was charged twice for order #12345.
The AI searches Stripe and identifies the duplicate charge. Then it creates a task in Asana for the sales team to process the refund.
The sales team member processes the refund and updates the task status in Asana. The AI monitors the task status and automatically notifies the customer once the refund is marked complete.
📅 Meeting assistant AI agent 🎯 Objective 🎯 j
Build an AI agent that manages a sales director’s daily schedule by analyzing meeting locations and travel logistics. The AI agent reviews calendar events, calculates travel times between locations, adds travel blocks automatically, and alerts the team about scheduling conflicts or impossible commutes.
🏗️ How to do it Connect your AI Agent to MCP servers and Make scenarios:
Google Calendar MCP server to read meeting schedules and add travel time blocks Google Maps API (via Make scenario) to calculate travel times and distances between meeting locations
🎬 See it in action The AI agent checks the Sales Director’s calendar each morning. Today’s schedule shows: 9 AM meeting in downtown office, 11 AM client meeting 25 miles away, and 12 PM lunch meeting back downtown.
The AI agent calculates travel times using Google Maps and identifies a problem: the 11 AM meeting location is 40 minutes away, making it impossible to attend the 12 PM meeting on time.
It sends a Slack alert: ⚠️Schedule conflict detected: Travel time between 11 AM and 12 PM meetings is 40 minutes, but only 1 hour available. Recommend rescheduling 12 PM meeting to 1 PM.
It also automatically adds a 40-minute travel block after the 11 AM meeting to the calendar.
How to use it
You can add MCP tools to your AI agent directly from within the agent scenario. To do so, hover over the + button of the Run an agent module and select Add tool. Then select the MCP Client> Call a tool module. Then you need to write a tool
name and description. This helps the agent understand what the module does and when to use it.
Next, you need to set up the MCP module by creating a connection to an MCP server.
There are three ways to do this.
Use an existing connection
Select one of the MCP servers you’ve previously connected to. Connect to verified MCP server
Select a server from Make’s verified list. Make has already configured the URL and connection details, so you just need to authenticate and start using it. Connect to a server that isn’t in the list
If your server isn’t in the verified list, you can connect to it by providing its URL. Select + New MCP server, enter the URL, and complete the authentication. After you’ve created the connection, select the tool you want to run and fill in the required information that the tool needs.
Note that you can set certain parameters to Let the AI agent decide. The module then uses whatever values the AI agent provides based on the current task.
See it in action
Here’s an example of using MCP tools in your AI agents. Build an AI Agent that receives cooking requests and
creates tasks in Asana to manage your kitchen operations.
The kitchen manager plans all kitchen work in an Asana project.
It stores all tasks related to cooking assignments.
Let’s have a look at how to set it up and see it working.
CREATE AI AGENT
Create an AI agent and set it up. ADD THE FIRST MCP TOOL
Hover over the + button of the Run an agent module and select Add tool. Then add the MCP Client> Call a tool module. Enter the tool name and description. Then connect to the Asana MCP server by selecting your existing connection. Select the Asana list workspaces tool and configure it. ADD THE OTHER MCP TOOL
Repeat the process to add the remaining tools: Asana get projects and Asana create task. Remember to choose Let the agent decide for fields the agent will provide. You can choose only the field you want your AI agent to provide. USE THE AI AGENT
Use the chat to ask the AI agent to create a new task in Asana and see that it uses the MCP tools to complete this task. Great! You’ve now seen how to add MCP tools to your AI agent and how it uses them to complete tasks. In this example, you had a simple, straightforward task, that’s why you used MCP tools.
MCP in Make scenarios
You can use the Make MCP client within your scenarios by adding the MCP Client> Execute an action with AI module.
When you connect to an MCP server with this module, all its tools become available to the AI. You can then select which ones your AI agent can access. The AI analyzes your request and automatically selects the right tools to complete the task.
Here’s how it works:
Step 1 Create a connection to your MCP server Step 2 Select which tools from that server the AI can access
Step 3 Describe what you want to accomplish in plain language
Step 4 The AI figures out which tools to use, in what order, and with what parameters
Step 5 The AI executes the tools and returns the results
This module is what you’ll focus on for the rest of the unit. You’ll see it in action soon.
When to use it
When should you use Execute an action with AI versus an AI agent?
Both let you describe a task and have AI select the right tools to complete it. The difference is the complexity of your task and how many services you need.
Execute an action with AI AI agents
It works with a single MCP server It is built for complex tasks across multiple It works with a single MCP server. It is built for complex tasks across multiple services.
Use it when you want the AI to perform They can use tools from different MCP actions using tools from one server. The AI servers and Make modules at the same reads your task description, picks the right time. tools from that server, and executes them in the right order. If your task involves getting Asana data, creating a PDF, and sending an email, use If you want to get all high-priority Asana the AI Agent to handle all these steps tasks and create a summary, use the across multiple applications. Execute an action with AI module
Real-life example
Let’s have a look at a real-life example.
👔 Client meeting preparation scenario 🎯 Objective Build a scenario that automatically prepares briefing documents before client meetings. The scenario gathers information from before client meetings. The scenario gathers information from your CRM and generates a comprehensive meeting brief with relevant client data and talking points.
🏗️ How to do it Build a scenario with:
Calendar module to detect upcoming meetings and extract client names
Execute an action with AI module connected to your CRM’s MCP server (with access to tools like get_account, get_deals, get_contacts, get_activities, update_account) Email module to deliver the briefing document
The AI automatically decides which CRM tools to call based on
🎬 See it in action You have a meeting with Acme Corp at 2 PM. At noon, the scenario detects this meeting and sends the client name to Execute an action with AI with the task: Prepare a meeting brief for Acme Corp including account status, active deals, recent
activities, and key contacts. If any opportunity value exceeds $100K, set the account priority level to High.
The AI calls multiple tools from the CRM MCP server:
get_account: finds company details get_deals: discovers $150K opportunity update_account: sets account to High priority since deal exceeds $100K get_activities: sees last contact was 2 weeks ago get_contacts: identifies decision-maker Then it generates a brief with this information.
You receive an email: Meeting Brief: Acme Corp - $150K deal in final stage, decision-maker is Jane Smith (CFO), last contact 2 weeks ago. Recommended: Follow up on pricing proposal and confirm timeline.
💜 Advantage of using Make The Execute an action with AI module automatically determines which CRM tools to call. You don’t need to map out every possible situation manually.
How to use it
To use the Make MCP client in your scenario, you first need to add the Execute an action with AI module. Next, select the MCP server you want to connect to and set up the connection.
You have the same options as before:
Existing connection Verified MCP server Add a new server
Select a connection you have Select one of the MCP servers Provide the URL to connect to a
previously created. verified by Make and authenticate server not present in the Make
to create a connection. verified MCP server list. Once you’ve connected to the MCP server, select which tools you want the AI to access and use in your scenario. Then specify what task you want to accomplish. When you run the scenario, the AI will select the tools it needs.
Let’s see how this works in practice.
See it in action
Here’s an example of using MCP tools in your scenarios.
Build a scenario that reads a task list and creates only new, non-duplicate tasks in Asana. A project manager uses the following tools to manage their work:
A Google spreadsheet where different teams add their task requests.
Asana projects to manage the different tasks.
They build a Make scenario to transfer and manage these tasks in Asana. The scenario retrieves new tasks from the Google Sheet and combines them into a single list. Then uses the Asana MCP server
tools (get_projects, get_tasks, create_tasks) to check if each task already exists, if not, it creates the task in the corresponding
project.
Let’s see it in action. Great! You’ve learned how to use the Make MCP client in scenarios. Now you’re ready to use it in your own projects.
Continue to complete this unit
Wrap up
The Make MCP client lets you connect to MCP servers and use their tools in your AI agents and scenarios. Instead of creating the connections or setting up the calls yourself, you can easily access outside applications and tools that aren’t already in Make. Once you connect to an MCP server, all its tools are ready to use.
Use the Make MCP client for simple, straightforward tasks, like creating tasks in a project management tool or fetching customer data. Use Make scenarios or modules instead when your task is complex, involves sensitive data, requires control over which inputs and outputs the AI Agent sees, or needs steps to execute in a specific order.
You can use the Make MCP client in two places: with your AI agents or within your Make scenarios.
For AI agents, you need to use the MCP Client> Call a tool module and select which tools your AI agent can access. For scenarios, you can use the MCP Client> Execute an action with AI module where you describe what you want to accomplish and the AI automatically picks which tools to use.
Well done! You now know how to use the Make MCP client.
In the next unit, you will learn more about the Make MCP server. Mark this task complete to continue to the next unit.