Unique AI Space — a step-by-step guide for administrators to create, configure, and share AI-powered chat spaces on the Unique AI platform.

What is Unique AI?

Unique AI is an agent built on a foundational loop pattern. It continuously evaluates what to do next — whether to respond, call a tool, or stop. The loop ends when the agent determines no further tool calls are needed.

Unique AI's loop-and-tools framework enables modular, adaptive behavior, making it a flexible foundation for collaborative AI. The agent loop drives decision-making by reflecting on user instructions, deciding on tool use, and either responding directly or invoking tools, ending when no further tool calls are needed.

Who is it for

Benefits

Ease of Use: Minimal typing and one-tap configurations for administrators.

Integrated AI Capabilities: Combines internal knowledge search, web search, and file interaction with powerful AI models for reasoning and response generation.

Customizability: Advanced settings for tailored configurations suited to your organization's needs.

Enhanced Collaboration: Suggested prompts and knowledge scope rules improve team productivity.

Step-by-Step Space Configuration Guide

Step 1: Create a Unique AI Chat Space

1. Navigate to Space Management on the side navigation bar.

2. Click Create Space.

3. Select Unique AI Chat.

Step 2: General Configuration

Name the Space and Add Instructions

Step 3: Configuration

Collapse the configuration by clicking on the arrow.

Select the AI Model

• Choose an AI model from the dropdown menu.

  • For additional models go to Language Model under Advanced Settings. Choose an AI model by entering the LanguageModelName into the text field. Refer to the first column of the Model Availability Matrix and copy the corresponding Model Name into the text field.

• If your model is not listed in the LLM Availability Matrix, add the AI model by typing theLanguageModelName of the Language Model Info object into the field AI Model. To define the full Language Model Info object please see Detailed Configuration.

Add Title/Greeting, Description and Chatbot Disclaimer

For end users, this information will be presented as follows:

Add Suggested Prompts

1. Tap the Add Prompt button.

2. Locate the Add Prompt Title * text field (mandatory).

3. Type your question or command. Example: Model the financial impact of a 10% increase in raw material costs.

4. The prompt appears in the list and becomes available in the prompt selector within the chat.

5. To delete a prompt: locate it in the Suggested Prompts list and click the X icon to its right. The prompt is immediately removed.

How to Delete an Existing Prompt

1. Locate the Prompt: Find the specific prompt you wish to remove from the list of suggested prompts.

2. Click the Delete Icon: To the right of the prompt's title, you will see an "X" icon. Click on it.

3. Confirmation: The prompt will be immediately removed from your list of "Suggested Prompts".

The Suggested Prompts will be available in the prompt list in the chat

Step 4: Configure Sources & Tools

In the tool configuration a suite of built-in tools designed to enhance the functionality and efficiency of Unique AI chat spaces is available. These tools are seamlessly integrated into the platform, providing users with powerful capabilities to interact with data and services.

Below are the key benefits of using these tools:

Internal Search

Enables RAG-based search across connected knowledge bases. When enabled with no specific scope defined, users can access all folders and files they have permission to view (default). Click the tile to access settings and connect a specific Knowledge Base. Additional details about knowledge base access can be found here: Knowledge Base for Admins.

Connect to Knowledge Base

To make only specific knowledge available for this Space, click on the button ‘Connect to a Knowledge Base’ and select the relevant folders or files in the following view. This ensures that users can only view the selected files within the created space.

Filter the Knowledge Base

You can enhance a space by adding smart rules, including Metadata Filtering. It allows users to perform detailed searches by filtering through metadata attributes like filenames, dates, and more. Details on how to configure this can be found here: Defining Smart Rules through Knowledge Scope Rules

Web search

Web Search is a built-in tool that enables the AI to access real-time information from the internet. This tool enhances the chatbot's capabilities by searching the web for current data, recent events, and up-to-date information that may not be available in the internal knowledge base.

Deep Research

Deep Research combines web and internal search with extended reasoning to produce in-depth, report-style outputs. Runs for 5–15 minutes and shows progress steps as it works.

Enable Uploaded content in Chat

Unique AI supports document and image uploads, allowing users to chat directly with uploaded files. When files are uploaded, the default behavior is that the model reads the full content of all uploaded files from the entire chat history on every conversation turn, if they fit within the context window of the model. This enables features like document comparison and summarization across multiple files.

This feature can be disabled in the Advanced Setting described in the Detailed Configuration if needed, particularly for long conversations where continuous reading of all historical uploads may impact performance.

Step 5: Sharing the Space

1. Use the Access button to add people or groups at the top right.

2. Assign permissions: "Can chat" or "Can manage."

Step 6: Publishing the Space

Review all configurations, then click Update to make the space live for your users.

Tips & Tricks

Use Suggested Prompts: Create ready-made conversation starters to guide users and improve first-time experience.

Leverage Advanced Settings: Customize AI models and tools for specific team or use-case needs.

Duplicate Spaces: Save time by duplicating existing configurations as the starting point for new spaces.

Limitations

• Context Window Internal Knowledge Search: The context window of the LLM defines how big of a “short-term memory” the model has for a single request.

• Deep Research Use Cases: Deep research use cases exceed current RAG capabilities but are planned for future implementation.