FinSight, is a conversational AI platform that transforms complex company financial documents (like 10-Ks, annual reports, and balance sheets) into clear, interactive insights. Built for investors, students, and analysts, FinSight personalizes your experience through a fun onboarding quiz and delivers everything from KPIs to educational mini-quizzes — all within a sleek mobile interface.
- Supports PDF uploads of financial documents
- Parses content using NLP to extract key metrics and trends
- Ask questions like:
- “What’s the EPS for this company?”
- “Summarize the risk factors from the annual report”
- Uses vector search + LLMs for contextual, document-aware answers
- Welcome quiz collects:
- Name
- Financial interests (e.g., investing, accounting)
- Skill level
- Sets tone and depth for insights (e.g., beginner vs. expert explanations)
- KPI cards show:
- Revenue, Profit, ROE, EPS
- Assets, Equity, Current Assets
- Visually organized with gradients, icons, and trends
- Users get quizzed on report data
- Incorrect answers include detailed explanations with references
- Gamifies learning about financials
- Start general chats without uploading
- Revisit and resume previous conversations
| Layer | Tech Used |
|---|---|
| Frontend | React Native (Expo), Recharts/Victory Charts |
| Backend | Node.js, Express.js |
| AI/NLP | Python (LangChain, PyMuPDF, OpenAI API, FAISS) |
| Database | MongoDB |
| Authentication | JWT, bcrypt |
| Hosting | Vercel (Frontend), Railway/Render (Backend) |
-
Clone the repo:
git clone https://github.com/your-username/FinSight.git cd FinSight -
Start the backend:
cd backend npm install npm run dev -
Start the frontend:
cd frontend npm install expo start
- Exportable PDF insights reports
- Multi-language document parsing
- Live valuation models
- Dhivyesh Prithiviraj, Kaushik Shivakumar, Adi Dixit, Kartik Emani