Skip to content

Watch it work, step by step

This is a replay of the assistant’s real step trace: what it is considering, which action it called, and the approval card that interrupts anything that spends money. It is a fixed script for illustration — it is not connected to a model, it takes no input, and it does nothing to any account.

Example runScripted replay — not a live model. Names and figures are illustrative.

Launch the Spring Intake campaign for the Toronto venture at $40 a day.

Done: Finding which brand that campaign belongs to.
Done: Checking your brands
Done: Pulling the campaign list for that brand.
Done: Pulling up your campaigns
Done: Refreshing live spend and results before changing anything.
Done: Refreshing live performance
Done: Launching spends real money, so this one stops here and waits for you.
Done: Preparing to launch the campaign
Done: Writing up the answer…
Approval needed

Launch “Spring Intake — Toronto” as a live paid Meta campaign at a $40.00 daily budget.

Approved — recorded, then run once with exactly these arguments.

Spring Intake — Toronto is live.

  • Daily budget: $40.00
  • Ad set, creative and ad were built from the first asset on the campaign
  • Status: ACTIVE

You approved this one, so it is recorded in the audit trail with your name and the exact arguments that ran. Want me to check the numbers tomorrow morning?