Put it into action
Objective
One AI Employee for a home-services client: it books jobs on a real calendar, saves every lead to the CRM, and checks the weather before it confirms outdoor work. You build it end to end, in a test account, one confirmed piece at a time.
Outcomes
✓You have a working AI Employee that books, captures leads, and checks the weather first
✓You can tell what the platform already does from what has to be built
✓You can test a tool where it will actually run, and read the Explanation view
This path covers
Three steps, in order: each one builds on the machine the last one left running. Booking and lead capture get switched on because the platform already does them; the weather check gets built because that judgment is specific to the client. Knowing which is which is the builder's skill this lab trains.
Switch on what the platform already does
Connect the calendar, switch on booking and lead capture, and watch a real booking land.
Key topics:
- The platform first, then the custom piece
- Switching it on
- Testing it, your first win
Build the custom tool
Turn a weather API into a tool, and write the prompt that gives it judgment.
Key topics:
- The job the platform does not already do
- Writing the capability prompt
- When to use it, and how to respond
Test it and run the whole thing
Test where it will actually run, read the Explanation view, and prove the full flow.
Key topics:
- Testing the tool where it will actually run
- Reading the Explanation view
- Running the whole flow end to end