> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cogniagent.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Self-scheduling
> Your AI coworker books its own follow-up — checking back on a deploy, waiting for a reply, or resuming a long job without you having to remember.
It books its own follow-up, so nothing waits on you remembering.
## What it does
Plenty of work isn't finished when you stop looking at it. An email needs a reply before the next step. A long process needs checking in an hour. A number needs watching until it crosses a line.
With self-scheduling on, your coworker can set its own reminder mid-task. It stops, comes back at the right moment, picks up where it left off, and tells you what happened while you were away.
You don't set anything up. It decides when coming back would be useful.
## When you'd use it
* "Check back in an hour and tell me if the deploy finished"
* Wait for a reply to an email it sent, then carry on from the answer
* Re-run a check every morning until a number crosses a threshold
* Pause on a long job overnight and resume without you
* Give a colleague until Friday to respond, then chase
* Watch a supplier's portal until the report appears
## Turning it on
Open your coworker, choose **Capabilities**, and find **Self-scheduling** in the Built-in tools list. It's on by default.
> **Self-scheduling** — Schedule its own follow-up turns to keep working or check back later.
## What it looks like
When your coworker comes back, a notice appears in the conversation saying it woke up and why. Everything it did while you weren't watching is there to read, in order, exactly as if you'd been sitting there.
If it finished the job, it says so. If it's still waiting, it tells you what for and when it'll look again.
## Three things that sound alike
They're easy to confuse, so here's the difference in one line each.
| | What it is | Who decides |
| ------------------------------------------------- | ------------------------------------------------------------ | ---------------------- |
| **Self-scheduling** | Your coworker comes back to a task it's already on | The coworker, mid-task |
| **[Schedules](/cowork/autonomy/schedules)** | Recurring work that starts on its own — every weekday at 9am | You, in advance |
| **[Keep working](/cowork/autonomy/keep-working)** | Keep going on this until it's done | You, in the moment |
## Good to know
Say what "done" means when you ask for something open-ended. "Check every hour until the report appears, then summarise it" gives your coworker a finish line. Without one it may keep checking longer than you meant.
A coworker that's waiting isn't using anything. It costs nothing to sit between check-ins — you're only paying for the moments it's actually working.
## Common questions
No. It works quietly and the result is waiting when you next look. The conversation shows what happened while you were away.
Yes. Open the task and stop it, and it won't come back. You can also just tell it to stop in the chat.
Far enough for anything task-shaped — hours, overnight, into next week. For work that should repeat indefinitely, use a [schedule](/cowork/autonomy/schedules) instead.
It gives up rather than checking forever, and tells you it did. You'll see the last thing it tried and why it stopped.
## Next steps
Recurring work you set up in advance.
Push through a long job in one run.
Everything that happens without you.
Wait on a person's reply.