> ## Documentation Index
> Fetch the complete documentation index at: https://docs.adopt.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Schedule and publish

> Set how often an agent runs and make it live for the people who use it.

Once an agent is configured and tested, you give it a cadence and publish it.

<Frame>
  <img src="https://mintcdn.com/adoptai-0ccbafe4/e2oYi85duHN1wflQ/images/acct-pipelines-library-1.png?fit=max&auto=format&n=e2oYi85duHN1wflQ&q=85&s=14e8e3de6c7d63ee62573fbfd55aa8ee" alt="Acct Pipelines Library" width="1495" height="812" data-path="images/acct-pipelines-library-1.png" />
</Frame>

## Choose a cadence

Cadence is set in plain terms — no cron expressions or pipeline wiring on the standard path:

* **Run once now** — a single run on demand.
* **Monthly** — good for recurring monthly deliverables.
* **Quarterly** — common for tax and review cycles.
* **At period close** — tie the run to the accounting period close.

Choose the rhythm the underlying work actually follows.

## Publish (go live)

Publishing makes the agent live: it runs on the schedule you set, prepares its work, and routes the output to the review queue for approval.

<Info>
  Publishing is safe — nothing an agent produces is shared before a reviewer approves it. Publishing starts *preparation*, not *delivery*.
</Info>

## Changing the schedule later

You can change cadence after publishing. If a client's needs shift (say, from quarterly to at-close), update the schedule; existing history is preserved.

## Advanced scheduling

On the Advanced path, scheduling is decoupled from the pipeline, so you get finer control over triggers if you need it. Most teams don't — the plain cadence options cover the common cases.

## Next steps

1. [Testing an agent](/builder-guide/testing-an-agent)
2. [Human-in-the-loop review](/core-concepts/human-in-the-loop)
3. [Notifications](/end-user-guide/notifications)