Workspace Studio
Workspace Studio is the management surface where your team authors everything an AI Assistant uses, beyond the tasks themselves. Four concepts live here, and they are best understood as a set rather than in isolation:
| Concept | What it shapes |
|---|---|
| Rules | Interpretation — what the assistant treats as noise vs signal, what to deprioritize, how to phrase findings. |
| Commands | Procedure — repeatable investigation flows; invoked via slash from chat or fired automatically on a cron schedule. |
| Knowledge | Context — narrative facts that telemetry cannot supply: ownership, maintenance windows, architecture assumptions. |
| Workflows | Triggers — event-driven automation: when an external event fires (alert, webhook, SLO violation), start an investigation and route the result somewhere. |
How to think about the four together
The first three — Rules, Commands, Knowledge — are the operational-context levers. They shape how an assistant reasons in your environment. None of them stand alone usefully:
- A Rule that says “deprioritize lab pressure” only matters when an assistant is interpreting findings.
- A Command that runs a checkout-flow investigation only matters once you have Rules to silence the lab noise it would otherwise trip over.
- A Knowledge note that says “the payments team owns checkoutservice” only matters if the assistant is investigating something connected to checkoutservice.
The fourth — Workflows — is the trigger surface. Workflows decide when an investigation starts in response to something happening outside RunWhen, and where the result is delivered. A workflow without good operational context just relays alerts; operational context without workflows still works interactively, just not autonomously.
A useful sequencing for adoption:
- Get tasks and SLXs producing useful Issues.
- Layer Rules so investigations read your environment correctly.
- Capture standing Knowledge facts that telemetry cannot supply.
- Encode repeated investigation flows as Commands.
- Wire those Commands into Workflows to fire on external events.
Most teams should not skip ahead — a workflow that fires an undertuned investigation produces noise on every alert, which is worse than no automation at all.
Where the lines fall
Some choices look like they could go in either of two places. The right answer is almost always:
| If you are encoding… | Use a… |
|---|---|
| ”This signal means X in our environment” | Rule |
| ”Always check these five things in this order” | Command |
| ”Here is a fact telemetry cannot tell you” | Knowledge note |
| ”When this external event happens, start an investigation” | Workflow |
When in doubt, the Building Operational Context guide walks through the four levers as a single workflow.
Where to use this in the UI
For the in-product Workspace Studio interface — tabs, forms, hierarchy panels, schedule editors — see Use → Workspace Studio.