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

# Run the pipeline

> Start the complete pipeline, begin at an advanced step, stop safely, and follow progress.

You can start the pipeline manually at any time from **Home**, or let a schedule trigger it automatically. Either way, the same eligibility rules, daily limits, and delivery modes apply.

## Run the complete pipeline

Select **Run Now** on the **Home** page to begin a full pipeline run. The system cascades through every eligible step for every active seat in this order:

<Steps>
  <Step title="Connection requests">
    Sends LinkedIn connection requests to contacts who are eligible and have not yet been invited.
  </Step>

  <Step title="Research and message preparation">
    Enriches contact data and generates personalised outreach messages ready for approval or automatic delivery.
  </Step>

  <Step title="Approved outreach">
    Delivers approved LinkedIn messages and emails that have not yet been sent, up to the seat's daily limit.
  </Step>

  <Step title="Due follow-ups">
    Processes due email and LinkedIn follow-ups for contacts who have not yet replied.
  </Step>
</Steps>

The run executes in the background — you can leave the page and monitor progress under **Activity**.

<Warning>
  Keep LinkedIn closed while the pipeline is running. Performing manual LinkedIn actions from the same account during the automation window can interfere with the run and trigger platform-level friction.
</Warning>

## Start at an advanced step

Open **Advanced** on the **Home** page when you want to skip earlier steps intentionally:

* **Messages** — start with research-complete prospects that still need message generation, then continue into outreach and follow-ups.
* **Outreach** — start with contacts whose messages are approved and unsent, then continue into follow-ups. Skips connection and research steps entirely.
* **Follow-ups** — process only due follow-ups for contacts who have already received initial outreach.

Starting at a later step does not change the eligibility of any contact. Every step still applies its normal gates, daily limits, and delivery mode checks. Use this option to catch up on a specific phase, not to bypass safety checks.

## Stop a run

Select **Stop** while a pipeline run is active. The current step finishes and saves its result before the run winds down — no action is left in an unknown or partial state.

<Warning>
  Stopping a run prevents future steps in that run from executing. It does not undo or recall anything that has already been sent.
</Warning>

The final status of every run appears in **Activity**:

| Status        | What it means                                                                                             |
| ------------- | --------------------------------------------------------------------------------------------------------- |
| **Succeeded** | The run reached normal completion. A succeeded run can still process zero items when nothing is eligible. |
| **Stopped**   | You stopped the run manually before all steps completed.                                                  |
| **Failed**    | An error prevented the run from completing normally.                                                      |
| **Skipped**   | A guard condition or empty queue prevented work in a specific step.                                       |

## Scheduled runs

When a seat's schedule is active, the pipeline starts automatically on the days and times you select in **Settings → Schedule**. The dashboard prevents two pipeline runs from overlapping — if a scheduled run starts while a manual run is still active, the scheduled run waits.

Manual **Run Now** is always available even when a schedule is configured. Use it to process urgent work outside normal hours without disabling the schedule.

For diagnosing runs that finish without producing the expected work, see [Activity and troubleshooting](/operate/activity-and-troubleshooting).