streamlit_demo

Streamlit Demo UI

A Streamlit-based user interface that can be used with any text-based Bedrock agent by updating the config.py file.

Using with Any Bedrock Agent

To add your own agent:

1. Add a new configuration to the bot_configs list in config.py:

{
    "bot_name": "Your Bot Name",  # Display name in the UI
    "agent_name": "your_agent_id", # Your Bedrock agent ID
    "start_prompt": "Initial message to show users",
    "session_attributes": {        # Optional: Include if your agent needs specific session attributes
        "sessionAttributes": {      # Custom key-value pairs for your agent's session
            "key1": "value1",
            "key2": "value2"
        },
        "promptSessionAttributes": {} # Additional prompt-specific attributes if needed
    }
}

Tested Demo Examples

The following demos have been tested with this UI and can be found in their respective folders:

• Sports Team Poet (/examples/multi_agent_collaboration/team_poems_agent/): Creates poems about sports teams
• Portfolio Assistant (/examples/multi_agent_collaboration/portfolio_assistant_agent/): Analyzes stock tickers
• Trip Planner (/examples/multi_agent_collaboration/trip_planner_agent/): Generates travel itineraries
• Voyage Virtuoso (/examples/multi_agent_collaboration/voyage_virtuoso_agent/): Provides exotic travel recommendations
• Mortgages Assistant (/examples/multi_agent_collaboration/mortgage_assistant/): Handles mortgage-related queries
• Custom Orchestration (/examples/agents/custom_orchestration_agent/): Demonstrates ReWoo (Reasoning without Observation) orchestration for a restaurant assistant agent

Prerequisites

1. Follow the setup instructions in each agent's respective folder before using them in the demo UI:

   • /examples/multi_agent_collaboration/mortgage_assistant/README.md
   • /examples/multi_agent_collaboration/voyage_virtuoso_agent/README.md
   • /examples/multi_agent_collaboration/trip_planner_agent/README.md
   • /examples/multi_agent_collaboration/team_poems_agent/README.md
   • /examples/agents/custom_orchestration_agent/README.md
   • /examples/multi_agent_collaboration/portfolio_assistant_agent/README.md

2. Ensure you have:

   • Python 3.x
   • AWS credentials configured with appropriate permissions

3. Create and activate a Python virtual environment:

   python3 -m venv .venv
   source .venv/bin/activate  

4. Install required dependencies:

   pip install -r src/requirements.txt

Running the Demo

3. Configure your AWS credentials with appropriate permissions

4. Run the Streamlit application:

   cd examples/agents_ux/streamlit_demo/; streamlit run demo_ui.py

5. Optionally, specify a specific bot using the BOT_NAME environment variable:

   BOT_NAME="<bot-name>" streamlit run demo_ui.py

   Supported BOT_NAME values:

   • "Portfolio Assistant"
   • "Sports Team Poet"
   • "Trip Planner"
   • "Voyage Virtuoso"
   • "Mortgages Assistant" (default)
   • "Custom Orchestration"

Usage

1. The UI will display the selected bot's interface (defaults to Mortgages Assistant if not specified)
2. Enter your query in the chat input field
3. The agent will:
   • Process your request
   • Show the collaboration between different agents
   • Display thought processes and tool usage
   • Provide a detailed response

Architecture

The demo UI integrates with Amazon Bedrock Agent Runtime for agent execution and showcases multi_agent_collaboration features including:

• Dynamic routing between specialized agents
• Knowledge base lookups
• Tool invocations
• Code interpretation capabilities

Below is an example of the demo UI in action, showing the Mortgages Assistant interface: