🏋️ AI Body Weight Assistant

An advanced, multi-agent AI assistant designed for the GenAI Academy Hackathon. It helps users achieve their fitness and weight management goals through personalized research, synthesis, and structured planning.

🌟 Overview

The Body Weight Assistant leverages a sophisticated Single-Orchestrator Architecture built on the Google Agent Development Kit (ADK). It coordinates specialized sub-agents to move users through a seamless journey from data collection to a finalized, actionable fitness guide.

🚀 Key Features

• Personalized Assessment: Collects 7 key fitness metrics including weight, target goals, activity levels, and dietary preferences.
• Automated Background Research: Uses Google Search to retrieve authoritative nutritional and exercise data tailored to individual metrics.
• Intelligent Synthesis: Synthesizes raw research into a comprehensive weekly meal and workout strategy.
• Safety Guardrails: Automatically audits plans for health risks (extreme calories, dangerous routines) with a feedback loop for revisions.
• Stateful Orchestration:
  • Resume Capability: Automatically detects where the user left off if a session is interrupted.
  • Human-in-the-Loop: Pauses for user approval after research synthesis before finalizing the guide.
• Empathetic Coaching: Delivers responses in a motivational "fitness coach" persona.

🛠️ Architecture

The system is powered by Gemini 2.5 Flash and organized into several modular components:

Component	Responsibility
Orchestrator (root_agent)	Manages the workflow, handles state transitions, and coordinates sub-agents.
input_form_agent	Ensures all required metrics are gathered accurately.
google_search_agent	Retrieves real-world benchmarks and data points.
research_agent	Synthesizes complex data into a personalized strategy.
guardrail_agent	Audits synthesized plans for safety and health risks.
response_formatter_agent	Polishes and formats the final plan for user consumption.

📂 Project Structure

body-weight-assistant/
├── body_weight_assistant/
│   ├── agent.py          # Root Orchestrator definition
│   ├── sub_agents.py     # Specialized agent implementations
│   ├── tools.py          # State management and utility tools
│   ├── models.py         # Pydantic data models
│   └── prompt.py         # Core Orchestrator logic and scenarios
├── tests/
│   └── test_agents.py    # Comprehensive test suite
├── agent_pattern.png     # Architecture diagram
└── pyproject.toml        # Dependency management

🏗️ Getting Started

Prerequisites

• Python 3.13+
• Google ADK installed and configured

Installation

# Clone the repository
git clone https://github.com/your-repo/body-weight-assistant.git
cd body-weight-assistant

# Install dependencies using uv or pip
uv sync

Running the Assistant

adk run body_weight_assistant

Running Tests

# Using the local virtual environment
export PYTHONPATH=$PYTHONPATH:.
./.venv/bin/pytest tests/test_agents.py

👤 Author

Danny Khant
📧 dannypmkhant@gmail.com

---

Built with ❤️ for the GenAI Academy Hackathon.