Everything on this page exists and works today. It’s just not something a first-time user needs — come back here once you’re comfortable with the basics in Start Here.

​
Acting as a client (platform administrators)

If you manage multiple client workspaces, the workspace switcher in the top bar (labeled “Platform mode” when nothing’s selected) lets you step into any client’s workspace to work exactly as if you were one of their users. [SCREENSHOT NEEDED: Workspace switcher dropdown open, showing a list of clients and “Return to platform mode”]
  • Click the switcher → pick a client → you’re now “Acting as [Client Name]” everywhere in the dashboard (Overview, Workflows, Content, Analytics all show that client’s data).
  • Click Return to platform mode to step back out.
This is how a platform administrator actually does hands-on work inside a client’s workspace, rather than only managing it from the outside.

​
Client management

Full detail: Clients. Covers creating workspaces, adding domains, and tuning each client’s default AI model and cost alert threshold.

​
Registry

Sidebar → Registry. Five tabs of shared platform configuration: Providers (which AI providers — Anthropic, OpenAI, Google, etc. — are enabled), Models, Tools, Plugins, and Feature flags. Most of this is look-up only for a regular user — the Providers tab is the only one with a real browsable list; the others are search-by-ID. Editing anything here requires platform-admin permissions. Full detail: Registry.

​
Workflow Builder

Sidebar → Workflow Builder. Lets you look up any registered workflow by its ID and see its full step graph visually, or as raw JSON. [SCREENSHOT NEEDED: Workflow Builder page showing a workflow’s step graph]
Editing or publishing a new workflow definition here is a platform-administrator, JSON-editing feature — it directly changes what a workflow does for every user in your workspace. This isn’t something to experiment with casually; a malformed definition is rejected by the server, but a valid but wrong one will run for real.
Viewing an existing workflow’s graph, on the other hand, is safe for anyone — it’s a good way to understand exactly what steps a workflow like full-content-pipeline actually runs, in order.

​
Agent Center detail

Beyond browsing (covered in Dashboard Features → Agents), platform administrators can activate/deactivate individual agents from the quick-action menu on each agent’s card — useful for temporarily disabling one agent without touching the workflows that use it.