Unit Introduction
This is the last unit of the course AI agents in Make.
After all of this theory, it’s time to see everything in action!
In this unit you will learn:
How to build an AI agent in Make
How to run your AI agent
Who stole the Very Precious Gem The clock is ticking for the thief!
Demo overview
Time to see a demo of how build your an AI agent with Make!
The first step is to decide what you want your AI agent to do. The AI agent in this example is a police detective agent that will help investigate
robberies.
The AI agent’s goal is to figure out who stole the Very Precious Gem using the alibis
you give it. If the alibis don’t provide enough information, the AI agent will instead write follow-up questions you can ask to collect more clues. However, if there’s enough evidence, the AI agent will identify the criminal and send an email
to the captain with all the details. To build this agent, you need all the elements you learned about in the previous section.
Information related to the task
You need to create a Google spreadsheet. One tab contains the list of alibis the AI agent can access, and the AI agent uses another tab in the same spreadsheet to write the follow-up questions.
Tools
Your AI agent needs two tools to carry out these actions:
Send an email
Write follow-up questions
Scenario You need a Make scenario to set up and run your AI agent. This scenario provides the AI agent with the list of alibis and includes the Run an Agent module that you will use to define your agent.
Let’s keep going!
Demo
Your AI agent needs access to specific data or to perform specific actions to be able to do its job.
You need to create a Google spreadsheet with two tabs:
Alibis: contains the list of alibis.
Follow-up questions: an empty tab in which the AI agent will write follow-up questions if needed. Here are the alibis.
Let’s have a look at how to build the agent.
Build the scenario Step 1
This scenario sets up and runs your AI agent. It also gives the agent the list of alibis from Google Sheets, ensuring it has all the information it needs before it starts.
The scenario is in scheduled mode. You can trigger it manually by clicking Run once.
Set up the AI agent Step 1
Step 1
The Run an agent module is where you configure your AI agent and give it the instructions it needs. Step 2
Step 2
First, connect to an AI provider so your agent can access an LLM. If you have a Pro, Teams or Enterprise plan, you can use your own provider or the one Make offers.
For this demo, use the Make AI provider that is available for all plans. Step 3
Step 3
Select the Large model, testing showed it produces the best results for this task. Step 4
Step 4
Next, write the Instructions that define your agent’s role. Step 5
Step 5
Write the Input with the specific task you want it to complete.
Add module tool Step 1
Step 1
Now give your agent the tools it needs, starting with the ability to send emails. Since Make has a module for this specific action, add it as a module tool.
To do so, hover over the + icon of the Run an agent module and select Add tool. Step 2
Step 2
Then select Email > Send an email to a team member. Step 3
Step 3
Set the Tool name and description so that the AI agent will use to understand what the tool does. Step 4
Step 4
Then choose yourself as the recipient and set Subject and Content to Let AI agent decide.
You can also add details to add more explanation for the agent. The agent will determine these values based on the specific task.
Add scenario tool Step 1
Step 1
Next, add the tool that lets the agent write follow-up questions. The agent provides all the questions together, so you need to split them into separate lines. Since this requires multiple actions, build a scenario tool.
To do so, hover over the + icon of the Run an agent module and select Add tool. Then select Scenarios > Call a scenario.
Provide the scenario description so the agent understands what it does. In the Scenario dropdown, select + Create a scenario. Step 2
Step 2
Name your scenario and define both input and output.
When you click Create a scenario, Make opens the scenario editor where you can build it. Notice that Make automatically adds the Start a scenario and Return output modules. Step 3
Step 3
Build the scenario based on what you want the tool to do.
Here, the agent provides questions in an array.
Use an Iterator to split them into separate bundles so each question goes into its own spreadsheet row.
Add a Router to process the Return output module at the end, after the scenario finishes processing all items. Without it, the scenario would stop prematurely since the Return output module ends the scenario run. Step 4
Step 4
Turn ON your scenario.
Note that if you’re on the free plan, you can only keep two scenarios active at a time. If you have more than two active scenarios, you will see an error when you try to add your tools. The AI agent is now ready and it has everything it needs to carry out its tasks.
Time to run it!
Not enough information
Click Run once to run the scenario. The AI agent couldn’t determine who committed the crime with the information you provided, so you didn’t receive any email with the crime report. Go to the follow-up_questions tab of the spreadsheet and see that the AI agent added questions to try to identify the criminal, as expected. Identifying the criminal
Let’s add more alibis that will allow the AI agent to identify the thief. This time, when you run the scenario, the AI agent could identify who stole the Very Precious Gem and sent a report to your email with all the information. It didn’t add any follow-up questions this time. There’s no escape for Tyler! 💡 Pro tip If you’re curious about how your AI agent worked, look at the execution steps and the Reasoning tab.
You can find them in the output bundle of the Run an agent module. They show all the information you gave to the AI agent and the steps it took to complete the task. Open them one by one and read what your AI agent did. You will also see which tool the AI agent called and the information it received from that tool.
All done! Now that you’ve seen how to build and run an AI agent in Make it’s time to put it into practice.
Go to the Make website and create a new scenario. Click on the lightbulb in the bottom right to open the Resources. Select the Build the agent one and follow the instructions to build your first agent!
Continuetothewrapupforthisunit
Wrap up
To build an AI agent in Make, create a scenario and add the Run an agent module. Connect to an AI provider, select an LLM, and write instructions
that define your agent’s role and task. Hover over the + icon of the Run an agent module and select Add tool: select Make modules for single actions or build scenario tools for complex tasks.
Running your AI agent means running the scenario that contains it. Set the scenario to scheduled mode to trigger it manually. Once running, your agent works independently, calling tools and making decisions. Check the execution steps and the Reasoning tab in the output bundle to see what your agent did.
Now build your own AI agent. Go to Make and create a new scenario. Click the lightbulb in the bottom right to open Resources and select Build an AI agent. Follow the instructions to configure your agent. This hands- on practice shows how everything works together.
Unit complete!
Well done!
You have seen how to build and run your first Make AI agent! And you discovered who
stole the Very Precious Gem. Time to check what you have learned!
Mark this task complete to continue to the next unit.