Scheduled-run predictability
Which routes can reliably back a scheduled run.
A schedule is only as good as the thing that's around to run it. Proliferate creates the scheduled run on time no matter what, but whether that run actually starts working depends on the run location you picked in automations.
Cloud runs are the predictable choice
Personal cloud and organization cloud automations don't depend on anything on your machine. When the schedule fires, Proliferate provisions a cloud workspace and starts the session directly. Your laptop can be closed, asleep, or on another network- it doesn't matter.
If you want a schedule you can actually trust to run overnight or over a weekend, point it at a cloud target.
Local runs need the desktop app open
A local automation runs on your machine, so it needs the desktop app open and its local executor available at the moment the schedule fires. If the app isn't running, the scheduled run doesn't disappear- it sits queued, and its status reads Queued, local executor not available yet until you open the app and it gets picked up.
Local automation run stuck queued
A run history row showing the 'Queued, local executor not available yet' status for a local workflow, next to a normal 'Session started' row.
That makes local automations fine for things you trigger manually or run while you're already working, but a weak fit for anything you need to fire on a schedule unattended. If predictability matters, switch the automation's run location to personal cloud instead- same prompt, same schedule, no dependency on your machine being awake.
Team automations are cloud-only by design (target_mode for an
organization-owned automation is always organization cloud), so this
local-vs-cloud tradeoff only comes up for personal automations.
What to check if a schedule looks unreliable
- Confirm the run location: local runs need the app open, cloud runs don't.
- Check the schedule itself. Today automations only support hourly or daily recurrence (daily can be limited to weekdays or weekends); anything more elaborate isn't supported yet, so an oddly-shaped schedule may not mean what you expect.
- Open the run history on the automation's detail page and look at the status of the last few runs before assuming the schedule itself is wrong.