What belongs in a workspace
- One or more AI products (Website, API, or Streaming API)
- Shared datasets and evaluation configurations
- Team members with role-based access controls
- Quota allocation and usage tracking
Workspace tiers
Workspaces are organized by tier, which determines available features and quotas.Roles and permissions
Each workspace member has one of four roles:Invitations
Invite team members by email with a specific role. Invitations can be:- Pending — Waiting for the recipient to respond
- Accepted — Member has joined the workspace
- Declined — Recipient declined the invitation
- Expired — Invitation was not accepted in time
Usage and quotas
Each workspace has a quota allocation that resets monthly. Track usage in the Usage & Limit page:- Total allowed evaluation runs per month
- Current usage with reset date
- Usage ledger showing per-run deductions
Workspace setup checklist
1
Name and scope the workspace
Match the scope to a product or team so results are actionable.
2
Invite the right people
Add product, ML, security, and compliance stakeholders. Assign roles appropriately.
3
Create your first product
Add a product (Website, API, or Streaming API) and follow the onboarding roadmap.
4
Add datasets
Browse the Dataset Marketplace or upload your own datasets.
Best practices
- Keep one workspace per product line or team to avoid mixing metrics
- Use the quota dashboard to monitor evaluation run usage
- Assign the minimum role needed — use Viewer for stakeholders who only need read access
Related docs
Get started
Launch your first workspace and evaluation.
Monitoring
Track production health per workspace.