Why this is different from a chat window
Real tools, not descriptions
The model calls the same endpoints your code would, and gets the same answers.
Runs on a schedule
Nobody has to be sitting there. The interesting output arrives on its own.
Billed once
The tool run is charged where it happens, once. An agent-driven run does not cost you twice for passing through two services.
Attributed properly
Usage shows which tool the agent reached for, not one opaque total.
Your key or ours
Bring your own provider key and pay a smaller platform rate, or use ours and pay one bill.
Capped
A monthly credit cap per job, because an agent that can call tools is an agent that can call a lot of them.
Give it a job and see what it reaches for
The run log shows every tool it called and what each one cost.
Set one up