GrowTheory - AI-Powered Company Intelligence Platform

An AI agent that provides real-time financial intelligence on publicly traded companies, helping investors and analysts make data-driven decisions.

Live Demo: growtheory.vercel.app

---

What It Does

GrowTheory is an autonomous AI agent that analyzes companies by:

• Gathering financial data from Yahoo Finance (revenue, market cap, profitability, stock performance)
• Analyzing market sentiment via AlphaVantage and NewsAPI
• Providing economic context using Federal Reserve economic indicators (FRED)
• Generating intelligence reports with investment outlook and risk assessment

Simply search for any publicly traded company ticker (e.g., AAPL, MSFT, TSLA) and get a comprehensive analysis in under 3 minutes.

---

Architecture

Tech Stack

• Frontend: React + Vite, deployed on Vercel
• Backend: AWS Lambda (Docker container)
• AI Agent: Amazon Bedrock (Claude 3.5 Haiku) + Strands SDK
• Data Sources:
  • Yahoo Finance (yfinance)
  • AlphaVantage API
  • NewsAPI
  • Federal Reserve Economic Data (FRED)
• Storage: DynamoDB with TTL-based expiration
• API Gateway: REST API with CORS-enabled endpoints

Agent Architecture

The AI agent uses multi-stage reasoning powered by Amazon Bedrock:

1. Financial Analysis Tool - Fetches real-time market data, calculates health scores
2. News Sentiment Tool - Analyzes recent news and market sentiment
3. Agent Orchestration - Claude autonomously decides which tools to call based on context
4. Report Generation - Synthesizes data into actionable intelligence

AWS Services Used

• Amazon Bedrock - LLM inference (Claude 3.5 Haiku)
• AWS Lambda - Containerized Python application with agent logic
• Amazon API Gateway - RESTful API endpoints
• Amazon DynamoDB - Company analysis cache with TTL
• Amazon ECR - Docker image registry
• Amazon CloudWatch - Logging and monitoring

---

Features

Intelligent Caching

• Multi-layer caching reduces latency and costs:
  • Frontend cache (5 minutes)
  • Lambda memory cache (5 minutes)
  • DynamoDB persistent cache (24 hours)

Smart Pagination

• Dashboard displays analyzed companies with pagination
• Real-time cache status tracking

Graceful Degradation

• Agent continues analysis even if individual APIs fail
• Transparent status reporting (complete, partial, failed)

Investment Intelligence

• Overall health score (0-100)
• Letter grade assessment (A+ to D)
• Green flags and risk factors
• Market sentiment analysis
• Investment outlook (BULLISH/NEUTRAL/BEARISH)

---

🎮 How to Use

1. Visit growtheory.vercel.app
2. Search for a company using its stock ticker (e.g., AAPL, MSFT, GOOGL)
3. Wait ~2-3 minutes while the AI agent gathers and analyzes data
4. Review the comprehensive intelligence report with:
   • Financial health score
   • Market sentiment
   • Investment outlook
   • Risk assessment

---

Project Structure

growtheory/
├── frontend/               # React application
│   ├── src/
│   │   ├── components/    # UI components
│   │   ├── pages/         # Page components
│   │   ├── services/      # API service layer
│   │   └── styles/        # CSS files
│   └── package.json
│
├── backend/
│   ├── agents/            # Strands AI agent
│   │   └── company_analyst.py
│   ├── tools/             # Agent tools
│   │   ├── financial_analyzer.py
│   │   └── news_analyzer.py
│   ├── lambdas/           # Lambda handler
│   │   └── lambda_handler.py
│   ├── Dockerfile         # Container definition
│   └── requirements.txt
│
└── README.md

---

🛠️ Local Development (Optional)

The application is fully deployed, but if you want to run it locally:

Prerequisites

• Python 3.13+
• Node.js 18+
• AWS Account with Bedrock access
• API Keys: AlphaVantage, NewsAPI, FRED

Backend Setup

cd backend
python -m venv .venv
.venv\Scripts\activate  # Windows
source .venv/bin/activate  # Mac/Linux

pip install -r requirements.txt

# Set environment variables
export ALPHAVANTAGE_API_KEY=your_key
export NEWS_API_KEY=your_key
export FRED_API_KEY=your_key
export COMPANY_CACHE_TABLE_NAME=your_dynamodb_table

Frontend Setup

cd frontend
npm install

# Create .env.local
echo "VITE_API_BASE_URL=your_api_gateway_url" > .env.local

npm run dev

---

Hackathon Compliance

AWS AI Agent Global Hackathon Requirements

✅ LLM hosted on AWS

• Amazon Bedrock with Claude 3.5 Haiku

✅ Agent Framework

• Strands SDK with custom tools

✅ Autonomous Capabilities

• Agent independently decides which tools to call
• Multi-stage reasoning with context awareness

✅ External Integrations

• 4 external APIs: Yahoo Finance, AlphaVantage, NewsAPI, FRED
• Real-time data fetching and synthesis

✅ AWS Services Used

• Amazon Bedrock (LLM)
• AWS Lambda (compute)
• Amazon DynamoDB (storage)
• Amazon API Gateway (API management)
• Amazon ECR (container registry)
• Amazon CloudWatch (monitoring)

---

Performance

• Average analysis time: 120-180 seconds (fresh data)
• Cached results: < 1 second
• Lambda cold start: ~1.5 seconds
• Lambda warm execution: ~10ms (with cache hit)
• DynamoDB reads: Minimal due to multi-layer caching

---

Security Features

• API Gateway rate limiting: 10 req/sec, 20 burst
• CORS-enabled endpoints
• IAM least-privilege policies
• Environment variable management
• No sensitive data in frontend

---

👥 Team

Built by Naimul Islam & Sayeed Ali

---

Acknowledgments

• AWS Bedrock team for Claude API access
• Strands SDK for agent framework
• AlphaVantage, NewsAPI, and FRED for data APIs