Standing up a weekly creative engine

The engine is assembled, not automatic

"A weekly creative engine" sounds like one switch you flip on. In Pinnora it is a few real pieces you connect by hand, and none of them generates, tests, and scales on its own. It helps to know what each piece actually does before you lean on it.

Autopilot rotates what you already made

Autopilot schedules creatives onto a recurring calendar. It does not make them. Its planner takes the artifacts you select and drops each one into a slot, using each exactly once. So the order of operations is: generate your creatives in the Studios first, then point Autopilot at the pile. Its real controls are a cadence, the days and platforms, and a standing budget cap that auto-pauses the automation when the cap is hit. There is no objective or KPI field, and it does not read performance to adjust what it posts.

It is a dishwasher on a timer. It washes what you load, on the schedule you set, and it never decides the menu or goes shopping.

Workflows are durable multi-step plans

Workflows are set up by telling Atlas a multi-step goal, not by filling in a form. Two shapes exist today: schedule a set of creatives you already have, or generate a set and then schedule them. They run durably and survive restarts, which starts to matter once you are scheduling many slots at once.

The honest limit

Neither Autopilot nor Workflows closes a learn-from-results loop, and in production no automations have run yet. What you can build is set-and-forget scheduling of creatives you still generate and choose yourself. That is genuinely useful for keeping a steady output going, as long as you do not expect it to improve the creatives based on how they perform.