🧠 CapitalAgent — Built for Beginners. Powered for Pros.

Timeless decisions for timeless returns
Built for WeHack 2025 | Theme: Timeless Moments Await

💡 Inspiration

Our mission is to make investing accessible for beginners while staying powerful and insightful for experienced professionals.
CapitalAgent is a conversational AI-powered investment platform that empowers users to evaluate and manage investments in two key asset classes:

📈 Stocks
🏠 Real Estate

Whether you're new to finance or a seasoned investor, we help you make smart, data-driven decisions with confidence.

🚀 What It Does

🧠 Advanced Risk Prediction
For professionals: input real-time stock or property data to generate detailed risk assessment reports.
👶 Beginner-Friendly Recommendations
For beginners: receive curated, low-risk suggestions in both domains, with explanations and trend forecasts.
📊 Personalized Portfolios
Each user gets a portfolio showing their investment history, performance, and current risk exposure.
⏳ Future Value Forecasting
Our models provide 5-year and 10-year predictions for both property and stock investments.
💬 Conversational Investment Chatbot
Powered by Gemini API, the bot explains risks in plain language and gives diversification advice based on your history (stored in MongoDB Atlas).

🏗️ How We Built It

🛠️ Backend: Python, Flask
📦 Database: MongoDB Atlas (to persist user investment history + queries)
🔮 AI/ML Models:
- LSTM for stock trend prediction
- Sentiment analysis for news insights
- Random Forest for real estate risk classification
🤖 Conversational AI: Google Gemini API (gemini-pro)
🧪 Testing: Postman, VSCode
🎨 UI/UX Design: Figma (UI mockups for future interface)

🧗‍♀️ Challenges We Ran Into

🔄 Google GenerativeAI version conflicts (v1beta vs v1) — had to carefully downgrade/upgrade to match the right model (gemini-pro)
🧩 Dependency errors — modules like sklearn, joblib, pymongo, and dotenv needed proper management in virtual environments
🧠 Contextual memory in chatbot — integrating MongoDB to persist user input + investment history required careful data modeling

🏆 Accomplishments That We're Proud Of

✅ Built a working end-to-end AI-powered advisor in under 15 hours
✅ Combined machine learning with LLM explanations
✅ Designed logic for investment diversification detection and advice
✅ Created a smart chatbot that learns from user behavior

📚 What We Learned

📦 How to structure real estate & stock data for AI modeling
🧠 How to integrate a large language model (Gemini) with memory (MongoDB) to simulate personalized financial coaching
⚙️ How to resolve API versioning and package conflicts across teams