# The Automation Ran Successfully. Nothing Got Done. — retained record

> Retained reddit record from reddit:AI_Agents.

Canonical: https://highsignal.app/data/records/4cbf81b780b34a1b

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:AI_Agents)  published 2026-09-26 11:54 ingested 2026-09-27 02:35

[original source ↗](https://www.reddit.com/r/AI_Agents/comments/1wqomu1/the_automation_ran_successfully_nothing_got_done/)

## retained text



```
One problem I kept running into with scheduled AI workflows was this: the automation would run, do some work, finish without an error and still leave the actual job incomplete. Say the target is 200 verified leads. If the automation finds 15 and stops, technically the run succeeded. But the workflow is still only at 15/200.

The mistake was treating each scheduled run as the job.

So, I changed the system. Now the target belongs to the workflow, not the individual run. A run is just another opportunity to move that target forward. The system also checks the actual output instead of trusting `last_run_time` or a “completed” status. If a run gets interrupted at 108/200, that progress is checkpointed and the next run continues from the live state.

I also removed arbitrary reasons to stop. No “batch complete”, no assumed 20-minute limit, no deciding it has done enough. If work remains and it can still work, it continues. I tested this on a research workflow. It survived interruptions, resumed from checkpoints and eventually reached exactly 200/200 before closing.

Turns out “Did the automation run?” is not a very useful question.

**“Is the work actually done?” is.**
```



document fetched 2026-09-27 00:19 · canonical https://reddit.com/r/AI_Agents/comments/1wqomu1/the_automation_ran_successfully_nothing_got_done
