Page	What You Can Do
Agents	Create, edit, favorite, and manage AI agents
Datasets	Upload documents for AI to reference
Teams	Share agents and datasets with teammates
Observability	View conversation history, token usage, and model details

Logging In

Open your web browser
Go to http://localhost:3002
Enter your credentials:
- Email: Your email address (created in Control Panel)
- Password: Your password
Click Sign In
If two-factor authentication is enabled, enter the 6-digit code from your authenticator app

First login? If you're using the default credentials (gtadmin@test.com / Test@123), go to Control Panel Guide first to set up properly.

Understanding the Interface

After logging in, you'll see:

Left Sidebar: Navigation tabs (Agents, Datasets, Teams, Observability)
Conversations List: Below the navigation tabs, shows your past conversations for quick access
User Profile: At the bottom of the sidebar, click your name to log out
Main Area: The content for the current page

Creating Your First Agent

An agent is an AI assistant configured with specific instructions, model, and optional knowledge (datasets).

Step 1: Go to the Agents Page

Click Agents in the left sidebar
You'll see two tabs:
- Favorite Agents: Shows your favorited agents for quick access
- Agent Configuration: Full list with all management options
Click Agent Configuration tab

Step 2: Create a New Agent

Click the Create Agent button (top right)
A form will appear with several fields

Step 3: Fill in Required Fields

Field	What to Enter	Example
Name	A descriptive name	"Customer Support Bot"
Model	Select from dropdown	"NVIDIA Llama 3.3 Nemotron Super 49B" (Requires NVIDIA NIM External AI Inference API Key
Category	Select existing or create new	"Support" or "Research"

Step 4: Configure Optional Settings

Field	What It Does	Recommended
Description	Explains what the agent does	Add a brief description
System Prompt	Instructions the AI follows	"You are a helpful customer support agent..."
Temperature	Creativity level (0-2)	0.7 for balanced responses
Visibility	Who can see this agent	Set based on your needs
Disclaimer	Warning text shown before chat	Optional legal/usage notice
Easy Buttons	Quick-action buttons for users	Pre-defined prompts users can click
Tags	Labels for organization	"support", "internal"
Datasets	Knowledge bases to reference	Select if you've created any

Step 5: Save the Agent

Review your settings
Click Save or Create
Your agent appears in the list

Verification: You should see your new agent in the Agent Configuration list with the correct name and model.

Chatting with an Agent

Starting a New Conversation

Click on an agent card from the Favorite Agents tab, or
Go to Agent Configuration tab and click on an agent to start chatting
The chat interface opens with an empty message area
Type your message in the text box at the bottom
Press Enter or click the Send button

Sending Messages

Type your message: Click in the input box and type
Send: Press Enter or click the Send button
New line: Press Shift+Enter to add a new line without sending
Wait: The agent's response will appear after a moment

Understanding Responses

The chat interface shows:

Response time: How long the AI took to respond (e.g., "1.2s")

Tip: To see detailed information like tokens used and model name, go to Observability in the sidebar and view the conversation details.

Message Actions

Hover over any message to see action buttons:

Copy: Copy the message text to clipboard
Download: Save the response as a file

Resuming Chat History

Your conversations are automatically saved. To continue a previous conversation:

Finding Past Conversations

Look at the Conversations List in the left sidebar (below the navigation tabs)
You'll see a list of your previous conversations
Conversations are sorted by most recent first

Opening a Conversation

Click on any conversation in the list
The conversation loads in the main chat area
You can see all previous messages
Type a new message to continue the conversation

Searching Conversations

If you have many conversations:

Use the search box in the conversations list area
Type keywords from the conversation
Matching conversations will be filtered

Favoriting Agents

Favorite agents appear in the Favorite Agents tab for quick access.

Adding Agents to Favorites

Go to Agents in the sidebar
Click the Favorite Agents tab
Click the Add Favorites button
A selection modal appears showing all available agents
Check the agents you want to add to favorites
Click Save or Done

Accessing Favorite Agents

Go to Agents in the sidebar
The Favorite Agents tab shows your favorited agents
Click any agent card to start chatting immediately

Removing from Favorites

Go to Agents → Favorite Agents tab
Click Add Favorites button
Uncheck the agents you want to remove
Click Save or Done

Creating a Dataset

Datasets are collections of documents that agents can reference when answering questions (called RAG - Retrieval Augmented Generation).

Step 1: Go to Datasets Page

Click Datasets in the left sidebar
You'll see any existing datasets and storage stats

Step 2: Create New Dataset

Click the New Dataset button
Fill in the form:
- Name: A descriptive name (e.g., "Product Documentation")
- Description: What this dataset contains (optional)
- Access Control:
  - Individual: Only you can use it
  - Team: Share with specific team members
  - Organization: Everyone in your organization can use it
Click Create

Verification: Your new dataset appears in the list with 0 documents.

Uploading Documents to a Dataset

Supported File Types

GT AI OS can process these file types:

Type	Extensions
Documents	PDF, DOCX, DOC
Text	TXT, MD (Markdown)
Data	JSON, CSV, XML
Web	HTML
Rich Text	RTF

Uploading Files

Go to Datasets in the sidebar
Find your dataset in the list and open it
Click Upload Documents button
Click Select Files button
Select one or more files from your computer
Click Upload

Understanding Processing Status

After upload, each document goes through processing:

Status	What It Means
Pending	Waiting to be processed
Processing	Currently being analyzed and chunked
Completed	Ready for AI to use
Failed	Something went wrong (check file format)

Processing typically takes a few seconds to a few minutes depending on file size.

Verification

After processing completes:

Document count increases
Chunk count shows how many pieces the document was split into
Storage size updates

Connecting a Dataset to an Agent

For an agent to use your dataset, you need to connect them:

Step 1: Edit the Agent

Go to Agents → Agent Configuration tab
Find the agent you want to update
Click the Edit button (pencil icon)

Step 2: Add Dataset

In the edit form, find the Dataset Selection section
Click to open the selection modal
Search or browse for your dataset(s)
Click on each dataset you want to connect
Click Save

How It Works

When you chat with this agent:

Your message is compared against the dataset
Relevant chunks are found
Those chunks are included with your message to the AI
The AI uses that context to give informed answers

Tip: Try asking the agent questions about the documents you uploaded to see RAG in action.

Exporting Agent Configuration

Export your agent settings to backup or share with others.

Step 1: Find the Agent

Go to Agents → Agent Configuration tab
Find the agent you want to export

Step 2: Export

Click the three-dot menu (⋮) on the agent row
Click Export
A JSON file will download to your computer
The file contains all agent settings:
- Name, description
- Model configuration
- System prompt
- Temperature, max tokens
- Tags
- Dataset connections

What's in the Export File

{
  "name": "Customer Support Bot",
  "model_id": "nvidia/llama-3.3-nemotron-super-49b-v1",
  "system_prompt": "You are a helpful...",
  "temperature": 0.7,
  "max_tokens": 4096,
  "tags": ["support", "internal"],
  ...
}

Importing Agent Configuration

Import a previously exported agent, one shared by a colleague, or a pre-built demo agent.

Step 1: Get the Import File

Use a JSON file exported from GT AI OS
Download a pre-built agent from Demo Agents
Or receive one from a colleague

Step 2: Import the Agent

Go to Agents → Agent Configuration tab
Click Import button (usually near Create Agent)
Click Browse or drag the JSON file
Review the agent settings that will be imported
Click Confirm or Import

After Import

The agent appears in your list
You may need to:
- Select a model (if the original model isn't available)
- Connect datasets (imported agents don't automatically connect to your datasets)

Managing Teams

Teams let you share agents and datasets with colleagues.

Creating a Team

Click Teams in the left sidebar
Click Create Team
Fill in:
- Name: Team name (e.g., "Engineering")
- Description: What the team is for
Click Create

Adding Team Members

Open your team (click on it)
Click Add Members or Manage
Enter email addresses of people to add
Click Add or Save

Note: Team members must already have accounts in GT AI OS.

When creating or editing agents/datasets:

Set Access Control to Team
Select which team(s) can access it

Tips for Better Results

Writing Good System Prompts

A system prompt tells the AI how to behave. Good system prompts:

You are a helpful technical support agent for [Company].
Your role is to:
- Answer questions about our products
- Help troubleshoot common issues
- Be polite and professional

Always:
- Admit when you don't know something
- Suggest contacting human support for complex issues
- Use clear, simple language

Organizing with Tags

Use tags to categorize agents:

By department: "engineering", "marketing", "hr"
By purpose: "support", "research", "creative"
By status: "production", "testing", "deprecated"

Common Issues

Go to Control Panel → API Keys
Verify you have at least one API key configured:
- NVIDIA API key for NVIDIA NIM models
- Groq API key for Groq models
Test each API key to ensure it's valid

"Agent doesn't seem to know about my documents"

Verify the dataset is connected to the agent (edit agent → check Datasets field)
Check document processing status is "Completed"
Try asking more specific questions about the document content

"Document stuck in 'Processing'"

Wait a few more minutes (large documents take longer)
If stuck for more than 10 minutes, try re-uploading
Check the file isn't corrupted or password-protected

"Chat input is disabled"

This can happen if:

No agent is selected (select an agent first)
Session expired (refresh the page and log in again)

"Conversation history is empty"

Conversations are stored per user account
If you logged in with different credentials, you won't see previous conversations
Check you're logged in with the correct account

What's Next?

Now that you know how to use the Tenant App:

Try the demo agents: Import pre-built agents from Demo Agents to get started quickly
Create specialized agents: Build agents for different use cases
Upload knowledge bases: Add your company documents to datasets
Share with your team: Use Teams to collaborate

Need help? Check the Troubleshooting guide.

Tenant App Guide

Table of Contents

Quick Reference