Finova

Inspiration

I wanted to build something that simplifies that process a single AI-powered personal finance assistant that not only tracks your expenses but actually helps you make smarter financial decisions.

What it does

Finova AI combines AI-driven insights, automated financial tracking, and a clean modern interface making it easier than ever to stay on top of your finances.

How we built it

-Frontend: React + TailwindCSS for a clean, responsive, and minimal UI. The interface includes interactive dashboards, financial summaries, and a conversational AI chat panel.

-Backend: FastAPI (Python) for a scalable REST API structure. It handles user authentication, receipt uploads, tax calculations, portfolio tracking, and AI-driven chat requests.

-Database/Search: Elasticsearch Cloud for intelligent data storage and retrieval. It indexes receipts, transactions, and financial summaries allowing semantic and hybrid searches.

-AI Layer: Google Vertex AI / OpenAI for natural language understanding and vector embeddings. This enables the chat assistant to provide personalized insights, tax advice, and portfolio optimization tips.

-Authentication: JWT-based token system for secure user sessions across frontend and backend.

Challenges we ran into

-Elastic Cloud Integration: Connecting FastAPI securely to a hosted Elasticsearch cluster required proper authentication setup, certificate handling, and managing network access.

-Proxy and CORS Issues: The React frontend initially failed to communicate with the FastAPI backend due to proxy mismatches (ECONNREFUSED and 401 Unauthorized errors). Debugging cross-origin policies and JWT validation taught us a lot about request flow and middleware logic.

-Asynchronous Indexing: Making Elasticsearch operations asynchronous (to prevent blocking FastAPI’s event loop) was tricky but necessary for scalability.

-AI Response Optimization: Getting the AI chat assistant to respond accurately to finance-related queries required prompt-tuning and handling structured data responses.

-Authentication Flow: Implementing token-based authentication and ensuring protected routes across all endpoints took multiple iterations.

Accomplishments that we're proud of

A working AI-powered finance assistant. Finova AI isn’t just a dashboard; it’s a tool that can think and analyze user data intelligently.

What we learned

How to connect FastAPI with Elasticsearch Cloud for smart, indexed data retrieval. Structuring a React + FastAPI full-stack app with proper token authentication (JWT). Creating a clean and intuitive dashboard using React, TailwindCSS, and charts. Using AI embeddings for semantic search and hybrid query matching in financial documents. Also,how to debug ECONNREFUSED and 401 Unauthorized like a pro 😅.

Built With

• anaconda
• elasticsearch
• javascript
• jwt/auth
• python
• react