Getting Started with Studio
Studio is an AI-powered data engineering platform that lets you build dbt models, semantic layers, and data validations through a conversational interface. This guide walks you through your first session.
Prerequisites
- An account for one of the Studio instance's configured sign-in Identity Providers
- Access to the Studio application URL
First-Time Login
Studio uses the sign-in-capable Identity Providers configured by the instance operator.
- Follow the Studio link from the website, or navigate to the Studio URL provided by your administrator.
- Choose a configured sign-in provider.
- Sign in with that provider.
- After successful authentication you are redirected back to Studio and land on the main workspace.
Note: If you reach Awaiting access, contact your administrator to confirm that your account is approved or that your Identity Provider has Auto-Approval enabled.
For the public Demo, the website sends you to Studio to sign in with GitHub. When the GitHub Identity Provider has Auto-Approval enabled, Studio creates and approves your User without assigning a platform role or Domain membership. You then use the Welcome Checklist to connect GitHub and provision your Demo.
The Welcome Checklist
After you sign in, the workspace landing screen remains visible and the Welcome checklist opens as a dialog over it. Studio refreshes the checklist whenever it opens, so shared setup completed by teammates and personal work you completed previously appear as Complete.
Shared readiness includes SSO, a Vibedata Owner, an LLM profile, the GitHub Commit Provider, and—for users with a role or Domain membership—an available Domain. If you are approved without either, Provision your demo replaces the shared Domain step and completes only after your own Demo Domain is active. Personal progress also includes your GitHub connection, an Intent you created, and your first attributable message. An action remains disabled when a prerequisite is missing or your role cannot perform it, with an explanation directly under that step.
Select an enabled Configure, Manage users, Connect, Create, Provision, New Intent, or Send a message action to continue setup. The checklist never takes you to another screen unless you select an action.
Select Close to dismiss the dialog temporarily. Finish setup remains in the bottom-right corner on the workspace and other screens and shows how many steps are left. Select it to open the same checklist over your current screen:
- Close closes only the current checklist window. Your progress is unchanged, and Finish setup remains available.
- Skip setup opens a confirmation. Confirming Skip setup permanently hides the checklist for your user on every device.
Steps that do not apply to the deployment's authentication mode do not block completion or offer an unusable action.
The checklist and Finish setup disappear automatically after all eight steps are complete or after you permanently skip setup.
Creating Your First Domain
Domains are top-level organizational groupings (for example, Sales, Finance, Procurement). You need at least one domain before creating an intent.
- Select Domains in the global application rail.
- Select Add Domain, or choose an example card to prefill an editable name, icon, and description.
- Enter a name (for example,
Sales Pipeline). - Select an icon that represents the domain.
- Optionally add a description, then choose the GitHub account and repository in Basics.
- Complete Data Platform, including Workspace, and Secret Store, then select Create Domain from any page. Domain Instructions can be added later in General.
The Domain appears in the persistent Domains list and becomes available in the sidebar's Domain Selector.
Creating Your First Intent
An intent represents a specific data question or pipeline you want to build (for example, "Quarterly Revenue by Region"). Each intent has its own conversation history and workspace.
- In the sidebar, ensure your domain is selected in the Domain Selector at the bottom.
- Click New Intent (the blue button at the top of the sidebar).
- A source selection dialog appears. Choose the data sources relevant to your question, then confirm.
- The workspace opens with an empty chat panel and "What would you like to build?" displayed in the center.
- Before sending your first message, use the AgentModelSelector in the bottom-left of the chat input to choose your preferred agent and model. The default agent is Data Engineer and the default model is claude-sonnet-4-5-20250929. The selection locks after your first message.
- Type your business question in the chat input and press Enter or click the send button (arrow up icon).
The agent begins working — you will see a typing indicator while it plans, then step-by-step progress in the right panel as it executes. If the agent needs clarification, it may present a QuestionCard wizard asking you to answer a few questions before proceeding (see Chat Interface for details).
The Three-Panel Layout
The main workspace is divided into three resizable panels:
| Panel | Purpose |
|---|---|
| Sidebar (left) | Lists all intents in the selected domain; domain selector at the bottom |
| Chat (center) | Your conversation with the AI agent; input bar at the bottom |
| Right Panel (right) | Scrolling Goal and context above pinned Usage and Current Viewers |
You can drag the resize handles between panels to adjust their widths. The sidebar can be collapsed entirely with Cmd+B / Ctrl+B.
Navigation
The top navigation bar provides access to the main views. Navigation is state-based — views replace the main content area without changing the browser URL (except for deep-link parameters described below).
| Icon | View |
|---|---|
| Home / Logo | workspace — the three-panel chat view |
| Code icon | code-view — file explorer and Monaco editor for the current intent |
| Eye icon | monitor — pipeline health dashboard |
| Domains | /domains — persistent Domain creation and configuration workspace |
| Gear icon | settings — source, skill, organization, and profile configuration |
Deep-Link URL Parameters
| Parameter | Effect |
|---|---|
?intent=<id> | Opens a specific intent directly on load |
?tab=<name> | Opens a specific settings tab |
?github_connected= | Shown after GitHub OAuth callback completes |
/domains/:id/:section | Opens a specific Domain section |
/domains/new | Opens inline Domain creation |
Quick Concepts
| Concept | Description |
|---|---|
| Domain | An organizational grouping for related intents (e.g., Sales, Finance) |
| Intent | A specific data question or pipeline build tracked as a conversation |
| Source | A connected data source (e.g., Salesforce, QuickBooks) the agent can query |
| Skill | An AI knowledge module that extends the agent's capabilities for a specific task |
| Artifact | A file produced by the agent (SQL model, YAML config, Markdown doc, etc.) |
When Studio Is Under Maintenance
Your administrator sometimes takes Studio offline for maintenance, most often to install an update. While that happens, every Studio page shows a dark Studio is under maintenance notice instead of the app.
- Your work is saved. Nothing you finished before the maintenance window is lost.
- A Studio tab you already had open moves to the same notice by itself, the next time it talks to Studio.
- You do not need to do anything. Reload the page to check if Studio is back.
- The notice disappears once your administrator confirms Studio is healthy again.
If the notice stays for longer than your administrator told you to expect, contact them — the notice stays up on purpose until they confirm Studio is healthy again.
What's Next
- Workspace Overview — Learn about the three-panel layout in depth
- Chat Interface — Understand all chat controls and interaction patterns
- Sidebar: Intents — Manage, rename, star, and organize your intents
- Domains — Configure Domains, Git repositories, and data destinations