# Start your first workspace > Coordinate related projects in dependency order with visible health gates and failure policy. Create a workspace in the browser after each member project is independently trusted and can reach its own success state. ## Review the graph Add projects and dependencies, then choose the partial-failure policy. Cycles are rejected. Independent branches may start in parallel; dependents wait for required health gates. ## Start from the CLI ```bash switchyard workspace list switchyard workspace get switchyard workspace start ``` The start command returns a durable operation. Open the workspace progress view to see queued, running, healthy, failed, skipped, and rollback states per project. ## Observable success Every dependency reaches its configured health gate before its dependents start. A partial failure follows the selected continue, stop, or rollback policy without tearing down project data. Bulk stop preserves volumes unless a separate destructive action is explicitly authorized.