UI Concepts Overview

The Cube AI user interface provides a web-based environment for interacting with large language models, managing domains, and performing user and domain-scoped operations.

This section provides both a conceptual overview and a visual walkthrough of the Cube AI UI.

UI Structure

The Cube AI UI is organized around a few core concepts:

Authentication – users log in using their credentials
Domains – isolated workspaces that define execution context
Navigation – domain-scoped features available through the sidebar
Actions – operations performed within the UI based on permissions

All interactions occur either:

Before selecting a domain (authentication phase), or
Within the context of an active domain (domain phase)

Authentication Flow

Users begin by accessing the Cube AI login screen.

Cube AI login screen

After successful authentication:

The user is redirected to the dashboard
Available domains are displayed
The user selects or creates a domain to continue

Account-related actions such as sign up, verification, password reset, and profile management are documented in:

👉 User Actions

Once a domain is opened, the UI switches into domain context.

Within this context:

All visible resources belong to the active domain
Available models and services are scoped to the domain
Visible actions depend on the user's assigned role

Switching domains updates the entire UI context accordingly.

Domain selection screen

For detailed domain workflows, see:

👉 Domains

Models and Inference

Models are scoped to the selected domain.

Users can interact with models directly from the Chat interface. The list of available models reflects what is enabled in the active domain.

Available models in the selected domain

Tokens (API Access)

The Tokens section allows users to:

Create Personal Access Tokens (PATs)
Define token duration and scopes
Revoke existing tokens
Copy tokens for API and tool usage

Tokens are required for API access and external integrations.

Personal Access Tokens screen

Chat Interface

The UI provides interactive access to language models through a chat interface.

Users can:

Select a model from available domain models
Send prompts and receive responses
Switch models dynamically
Execute inference inside a Trusted Execution Environment (TEE), if enabled

Responses are domain-isolated and follow the security policies of the active domain.

Chat and model selection screen

Roles and Permissions

Access to UI features is controlled by role-based permissions.

Depending on their assigned role within a domain, users may be able to:

Create or manage domains
Invite or manage other users
Access administrative features
View audit logs
Perform secure model operations

For details, see:

👉 Roles and Access Control

Security and Auditing

Cube AI provides visibility into security-relevant actions through audit logs.

Audit logs help track:

User activity
Domain-level changes
Role and membership updates
Secure model interactions (including TEE-based workloads)

For more details:

👉 Audit Logs

Typical User Workflow

A common end-to-end workflow looks like this:

Log in to the UI
Select or create a domain
Select an available model
Create a Personal Access Token (if API access is needed)
Interact with the model via Chat or API

Developer-Focused UI Documentation

This section focuses on user-facing UI behavior.

Developer-oriented documentation related to UI integration and customization, including the Chat UI and backend configuration, is available in:

👉 Chat UI

Next Steps

To continue exploring Cube AI:

Start with Domains
Review User Actions
Explore the Security & Access section