Runnit Studio
Runnit Studio is the in-app place to build custom artifacts with the AI builder agent. It is one feature with one storage and runtime model: the type of artifact you build determines where it surfaces in the app.
Artifact types
Section titled “Artifact types”| Type | Status | What it is | Where it appears |
|---|---|---|---|
| Page | Shipped | A full-screen custom page that owns the content area. | Its own route, granted into the sidebar. |
| Widget | Shipped | A compact tile sized for a dashboard grid cell. | The configurable dashboard catalog. |
| Form | Shipped | A Studio-native intake form with an attached workflow. | The Studio artifacts list. |
| Automation | Shipped | A trigger and a set of actions that run on their own, on an event, a schedule, an inbound webhook, or on demand. | The Studio artifacts list. |
| Block | Planned | A small reusable UI/logic piece embedded in other artifacts. | Coming soon. |
Publishing a widget to a dashboard
Section titled “Publishing a widget to a dashboard”- Create a widget in Studio (New → Widget → template).
- Build it manually or with the AI builder agent. It produces compact, size-aware output.
- Publish it organisation-visible from the Sharing tab.
- The dashboard’s widget picker lists it, and your placement persists per user.
Design system inside artifacts
Section titled “Design system inside artifacts”Studio artifacts render inside a secure sandbox that still looks native to Runnit. The EK Modena fonts are inlined, and theme-aware colour tokens are exposed as CSS variables so components match light and dark mode automatically. Authors style with these tokens rather than hard-coded colours.
When a user changes theme, published widgets receive the new tokens and recolour in place. The widget is not reloaded, so interactive state is preserved and there is no refresh flash.
Runtime AI in pages and widgets
Section titled “Runtime AI in pages and widgets”Published pages and widgets can use the workspace AI while people view them: generating summaries of the data they show, interpreting results, or embedding a chat with one of your organisation’s AI agents directly in the page.
This is off by default. An organisation administrator turns it on from the Studio settings page with Enable runtime AI for pages and widgets, and can set a daily runtime AI call budget that caps how many AI calls published components can make per day across the organisation.
Every runtime AI call is recorded on the AI Usage page with its token counts and cost, attributed to the component and the viewing user. Agent conversations started from a page run as the viewing user with their normal permissions.
See Component Builder for how the AI builder agent authors these artifacts, and Workflow Automations for building automations that run on a trigger.