Managing Workspaces

Once a workspace is created, you can control its lifecycle, update its configuration, and monitor its status from the dashboard. This guide covers day-to-day workspace management.

Starting, Stopping, and Restarting

Starting a Workspace

Workspaces start automatically when you open them (see Scale-to-Zero). You can also start a stopped workspace manually:

1. Go to the Dashboard.
2. Find the workspace card showing a Stopped status.
3. Click the Start button (▶).

The workspace pod is provisioned on the cluster and transitions to Running within a few seconds.

Stopping a Workspace

To manually stop a running workspace and deallocate its pod:

1. Open the workspace or find it on the Dashboard.
2. Open Workspace Settings → General.
3. Click Stop Workspace.

The workspace enters Stopped state. Conversation history, files, and all configuration are preserved — only the compute resources are released.

tip

Manually stopping workspaces is useful when you know a workspace won't be needed for a while. Otherwise, scale-to-zero handles this automatically after 30 minutes of inactivity.

Restarting a Workspace

To restart a running workspace (useful if something seems stuck or you've changed configuration that requires a fresh pod):

1. Open Workspace Settings → General.
2. Click Restart Workspace.

The existing pod is terminated and a new one is provisioned. This is equivalent to stopping and then starting the workspace. Conversation history is preserved.

Editing Workspace Settings

You can change a workspace's configuration at any time. Some changes take effect immediately; others require a restart.

Changeable Settings

Setting	Effect	Restart Required?
Workspace Name	Updates the display name on the dashboard and in the header.	No
AI Provider	Switches the AI backend (e.g., from OpenAI to Anthropic).	Yes
Model	Changes the model within the current provider (e.g., gpt-4o → gpt-4o-mini).	No
System Prompt	Updates the instructions sent to the AI at the start of each conversation.	No

How to Edit

1. Open the workspace.
2. Click Workspace Settings (gear icon).
3. Make your changes.
4. Click Save.

If a restart is required, you'll see a prompt. The workspace will restart automatically when you confirm.

:::warning Switching providers Changing the AI provider requires valid credentials for the new provider. Make sure you've configured the provider's API key or service account in your organization connections before switching. :::

Deleting a Workspace

Deleting a workspace permanently removes it and all associated data.

1. Open Workspace Settings → General.
2. Scroll to the Danger Zone section.
3. Click Delete Workspace.
4. Confirm by typing the workspace name.

:::danger This action is irreversible Deleting a workspace permanently removes:

• All conversation history
• All files and uploads
• All workspace-specific configuration
• The workspace's Kubernetes pod and storage

This cannot be undone. Export any data you need before deleting. :::

Workspace Status Indicators

The dashboard shows a color-coded status badge for each workspace:

Status	Indicator	Meaning
Running	🟢 Green	Workspace is active and ready for use.
Starting	🟡 Yellow	Pod is being provisioned. Usually takes a few seconds.
Standby	🟡 Yellow	Workspace is preparing to scale down due to inactivity.
Stopped	⚪ Gray	Workspace is fully stopped. No resources consumed.
Error	🔴 Red	Something went wrong during provisioning or runtime. Check logs.

If a workspace is stuck in Error state, try restarting it. If the issue persists, check that your AI provider credentials are valid and that your organization hasn't exceeded its resource limits.

Resource Limits

Each workspace pod runs with defined resource boundaries to ensure fair usage across the cluster and predictable performance.

Resource	Request (guaranteed)	Limit (maximum)
CPU	250m (0.25 cores)	1000m (1 core)
Memory	512Mi	2Gi

What This Means

• CPU: Your workspace is guaranteed 0.25 CPU cores and can burst up to 1 full core when capacity is available. This is sufficient for most AI interactions, code execution, and tool usage.
• Memory: Your workspace is guaranteed 512 MiB of RAM and can use up to 2 GiB. This covers chat history, file operations, and code execution with comfortable headroom.

info

Resource limits apply per workspace. If you have three running workspaces, each gets its own independent allocation. Stopped workspaces consume no resources.

When You Hit Limits

If your workspace exceeds its memory limit, the pod may be restarted (OOMKilled). If you're running memory-intensive operations (large file processing, heavy computation), keep an eye on usage.

If you consistently need more resources, contact us about Enterprise plans with custom resource allocations.