Skip to main content
To make the most of the UBIK API, it’s helpful to understand the key concepts that you’ll be working with. This guide introduces the core components of our platform.

Workspaces

High-level containers for organizing your projects and data.

Documents

The raw information you provide, like PDFs, text files, or web pages.

Tools

Powerful, pre-built functions that perform complex tasks on your documents.

Agent Sessions

Stateful conversations with AI agents that maintain context and history.

​
Workspaces

Think of a Workspace as a high-level container for your projects. It’s the primary way to organize and segregate your data. Each workspace can contain its own set of documents, tools, and assistants.
  • Organization: Group related documents for a specific project, team, or customer.
  • Isolation: Data within one workspace is kept separate from others, ensuring privacy and control.
  • Default Workspace: Every user has a default workspace where new documents are added if no other workspace is specified.
You can manage your workspaces through the /workspaces endpoints.

​
Documents

A Document represents any piece of content you ingest into the UBIK platform. This could be a file you upload (like a PDF or .txt file) or a web page you scrape. Once ingested, documents are processed and stored, making them ready to be used by Tools.
You can manage your documents through the /documents endpoints.

​
Tools

A Tool is a specialized function that performs a specific, complex task. Tools are designed to be powerful and asynchronous, allowing you to execute long-running jobs without blocking your application. Examples include:
  • Extracting financial data from a report.
  • Summarizing a long legal document.
  • Answering questions based on the content of multiple documents.
You can discover and execute tools using the /tools endpoints.

​
Agent Sessions

An Agent Session is a persistent, stateful conversation with an AI agent. Unlike simple chat completions, sessions maintain:
  • Conversation History: The full context of the interaction, including branching paths.
  • Document Context: Files and knowledge specifically linked to the conversation.
  • Tool Access: The set of tools the agent can use to answer questions or perform tasks.
Sessions are designed for building interactive applications like customer support bots, research assistants, or data analysis tools.
You can create and manage sessions through the /agent-sessions endpoints.