Platform Overview

The workspace selector determines the context in which you view and manage resources.

• A Personal space is for an individual user's resources.
• An Organization workspace is shared by organization members.
• Organization roles are Creator, Collaborator, and User.
• Resource-level access can grant additional access to a specific Customize Model, knowledge base, or skill.

A workspace is therefore not always an organization: it can be either a personal space or an organization.

A Customize Model is an API-callable model created from a base model and configured with the capabilities your application needs.

The six main configuration stations are:

• Overview — model identity, status, and integration examples
• Data — connected sources, knowledge bases, and SQL databases
• Business logic — business objects, vocabulary, and confirmed metrics across sources
• Behavior — Kernel Prompt, Agent Loop, skills, MCP, web search, API tools, image understanding, thinking, memory, and guardrails
• Evaluate — request-log evaluation and knowledge-base retrieval tests
• Schedule — self-trigger and bound SQL synchronization schedules
• Operate — requests, guardrail records, and integration examples
• Evaluate — evaluations, retrieval tests, and conversations
• Operate — request logs, guardrail intercepts, and integration examples

Settings contains model, access, and deletion controls. API key rate limits are managed on the API keys page, not in Customize Model settings.

A Knowledge Base stores documents that a Customize Model can retrieve as grounded context.

• Upload supported documents and images
• Parse scanned or visual content with the configured document pipeline
• Index document content for keyword, semantic, and hybrid retrieval
• Connect the knowledge base to a Customize Model from its Data station

UnieAI Studio exposes OpenAI-compatible endpoints through the LLM Runtime.

• User API keys are account-level credentials and are visible across personal workspace contexts.
• Organization keys are project-scoped credentials created and managed in an organization workspace.
• The runtime authenticates the key, checks model access, routes the request, records usage, and invokes configured model capabilities.

When migrating an existing client, configure the UnieAI base URL and API key, and set model to the base-model or Customize Model slug you want to call.