Step 3. Create a scenario to send tasks to the AI agent
This page reflects a previous version of Make AI Agents. For the latest information, see Make AI Agent (New).
In this final step, we create the scenario to communicate with our agent. We will equip the agent with scenarios as tools to perform the agent's tasks: list our shop inventory and order more items. We will use Slack to send tasks to our agent.
Click the Create a new scenario button in your organization dashboard or in the list of scenarios.
Add the Slack > Watch new events module to the scenario.
Set up the module:
- In the Webhook field, select a Slack event webhook or click Add to create a new one.
- In the Event type drop-down, select New channel message.
- In the Connection field, select your Slack connection.
- In the Channel field, select a channel you can use for testing the AI agent.
- Set the scenario scheduling to Immediately when data arrives.
- Click Save to confirm the module settings.
Click the module placeholder to add a new module to your scenario. Search "AI agent" to get the Make AI Agents > Run an agent module and add it to your scenario.
In the Agent field, select the agent you created in the previous section.
In Tools, click Add tool.
Add the scenarios to the AI agent's tools:
- Click the empty drop-down under Select scenario. The list of scenarios in your team rolls down. Select the "List shop inventory" scenario we created in the previous section.
Use the search row at the top of the selection list to find the scenario. Note the scenario description that appears under the name of the scenario and the name of the scenario folder.
- Click OK to confirm the selection.
- Add the "Create buy stock order" scenario created in the previous section in the same way.
- Click Save.
Finish setting up the Run an agent module:
- In the Thread ID field, map a unique identifier to keep the same thread for all communication with the agent. This optimizes the context sent to the agent with each module run and allows the agent to keep the history of previous conversations.
- In the Message 1 field, map the Text from the Watch new events module.
- Click Save to confirm changes.
Add the Ignore error handler to the Run an agent module. Sometimes, the agent can take time to respond.
If the response time gets longer than 180 seconds, the module outputs the ModuleTimeoutError and Make stops the scenario.
There are more sophisticated options to prevent the ModuleTimeoutError, but they are not part of this use case. If you want to learn more, check the module settings reference.
Add the Slack > Create a message module after the Run an agent module.
Set up the module:
- In the Connection field, select your connection.
- In the Enter a channel ID or name and Channel type fields, select values relevant for the channel where you want to send agent's replies and pick the channel in the next field.
- In the Text field, map the response from the AI agent.
- Click Save to confirm changes.
Update the name of your scenario.
Save the scenario.
You should now test your agent. Send a message to the channel that is watched by the Watch new events module that contains the intent for the agent to either send the inventory information or order new stock. For example, request the agent to list the current shop inventory.
Check the channel where is the Create a message module sending agent's replies. You should see a reply from the agent listing the same data as we defined in the data store when we were setting up the scenarios available to the agent.