See every background routine and scheduled AI job in one place, with its last result and a button to run it again. Change schedule or model without touching code, and the change applies immediately. A crashed run never blocks anything, it gets cleaned up at the next start.
The Agentic Part
An agent prevents accidental double starts by giving every run its own ID. If a run crashes, it cleans it up by itself at the next start, and a new routine registers itself in the registry the first time it runs, without anyone adding it by hand.
Main Features
- Unique run ID per job run, preventing accidental double starts
- Automatic cleanup of a job stuck after a crash
- Schedule changes take effect immediately, no restart
- New routines register themselves on first run
- Live progress visible after leaving and returning to the page
Use Cases
- Checking which AI job last ran and what it cost used to mean digging through several scripts. Now it sits in one place for every scheduled job, with its last result and a button to run it again.
- A job crashes in the middle of the night. At the next scheduled start, the agent cleans it up itself, nobody has to step in by hand.
- A new routine joins the estate. It registers itself in the registry the first time it runs.
- A schedule needs to change on short notice. The change applies immediately, with no restart of the application.
What we build for you
- Jobs and routines included, and what each writes or changes
- Schedule names and their default timing
- Event triggers in addition to schedule-based runs
