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

# Schedule agent runs

> Run recurring research, reporting, monitoring, and creative planning on a schedule.

Scheduled runs turn Superscale from a chat surface into an operating cadence. Instead of asking manually every time, schedule the agent to run the same job on a cadence and produce the output you need.

Scheduled agents run autonomously. That means the objective must be detailed enough for the agent to complete without asking follow-up questions at runtime.

## What to schedule

<CardGroup cols={2}>
  <Card title="Weekly competitor monitoring">
    Find new competitor ads, repeated formats, and emerging hooks every Monday.
  </Card>

  <Card title="Ad account health checks">
    Summarize winners, waste, fatigue, and recommended actions from connected accounts.
  </Card>

  <Card title="Creative refreshes">
    Generate new concepts based on recent winners, account data, and competitor movement.
  </Card>

  <Card title="Reporting">
    Send a concise report to the team: what changed, what to do, and what to make next.
  </Card>
</CardGroup>

## Set up a useful run

<Steps>
  <Step title="Name the job" icon="tag">
    Use a clear name: "Monday Meta health check", "Weekly competitor scan", or "Friday creative refresh".
  </Step>

  <Step title="Choose the sources" icon="plug">
    Pick the brand, products, competitors, ad accounts, reports, or context files the agent should use.
  </Step>

  <Step title="Define the output" icon="list-checks">
    Say exactly what you want: action list, report, five concepts, winners/losers table, or new briefs.
  </Step>

  <Step title="Set the cadence" icon="calendar-days">
    Use a one-time run for a specific future task, or a recurring run for daily, weekly, monthly, or custom cadences.
  </Step>

  <Step title="Attach skills if needed" icon="brain-circuit">
    Use selected [custom skills](/features/custom-skills) when the schedule should follow a repeatable method.
  </Step>
</Steps>

## Scheduling rules

| Rule                           | Why it matters                                                                                                                   |
| ------------------------------ | -------------------------------------------------------------------------------------------------------------------------------- |
| **One-time or recurring**      | A schedule uses either a future run time or a recurring cadence, not both.                                                       |
| **Local timezone**             | Times are interpreted in the user's local timezone from the browser.                                                             |
| **Self-contained objective**   | The scheduled agent cannot stop and ask clarifying questions when it runs.                                                       |
| **Integration checks**         | Schedules that depend on Meta, TikTok, Google, Slack, Shopify, or other tools need those integrations connected first.           |
| **Approval gates still apply** | A schedule can prepare recommendations, reports, and drafts, but approval-gated ad-platform writes still wait for your approval. |
| **Active schedule limit**      | Workspaces have a limit on active schedules, currently 10 per organization.                                                      |
| **Access gating**              | Scheduled runs may depend on plan or feature access.                                                                             |

## Better scheduled prompts

* "Every Monday, find five new competitor ads worth learning from and explain which format or hook we should test."
* "Every weekday at 8am, show underperforming ads to pause, winners to scale, and one creative concept to generate next."
* "Every Friday, compare this week's top creatives with our saved winners and suggest the next batch."

## Keep it narrow

Scheduled tasks should be specific. A narrow recurring job produces better action than a broad report that tries to cover everything.

<Card title="Performance monitoring" href="/features/performance-monitoring" horizontal>
  Turn scheduled analysis into decisions.
</Card>