# Experiments (/docs/guides/developer-workspace/pipelines/experiments)



An experiment is a named workspace that groups related pipeline runs
together. Use experiments to separate runs by project, model version, or
any other logical boundary. Every run must belong to an experiment, so you
need to create at least one before you can start a run.

Create an experiment [#create-an-experiment]

1. Select **Pipelines** in the left sidebar, then select the
   **Experiments** tab.

2. Select <kbd>+ Create experiment</kbd>.

3. Enter a **Name** and optional **Description** for the experiment.

4. Select <kbd>Next</kbd>.

   BullSequana AI creates the experiment and prompts you to start a run
   immediately. Select <kbd>Skip</kbd> if you want to start a run later.

The experiment appears in the **Experiments** list.

View an experiment [#view-an-experiment]

Select an experiment in the list to open it. The experiment detail page
shows all runs associated with it, along with their status, duration, and
start time. You can compare runs side by side to see how parameter changes
affect results.

Archive an experiment [#archive-an-experiment]

Archiving an experiment hides it from the active **Experiments** list
without deleting it. Archived experiments and their runs remain accessible
from the **Archive** view.

To archive an experiment, select the overflow menu (&#x2A;*⋮**) next to the
experiment and select **Archive**.

<Callout type="info" title="Archiving and runs">
  Archiving an experiment also archives all runs within it. You can restore
  the experiment and its runs at any time from the **Archive** view.
</Callout>

Next steps [#next-steps]

* [Start a run](runs) within an experiment.
* [Configure a recurring run](recurring-runs) to repeat a pipeline on a
  schedule.