Here's a hackathon submission template for ExpenseTracker Pro:
Inspiration
Managing expenses is tedious and time-consuming. We wanted to create a voice-first AI assistant that makes financial tracking as easy as having a conversation with a smart friend.
What it does
ExpenseTracker Pro combines voice AI, OCR receipt scanning, and Gmail integration to automatically track expenses. Users can ask natural questions like "How much did I spend on food?" and get intelligent insights with optional audio responses.
How we built it
- Frontend: React + TypeScript with voice recording capabilities
- Backend: Node.js/Express with OpenAI Whisper for speech-to-text
- AI: GPT-4 for expense insights and text-to-speech for audio responses
- OCR: OpenAI Vision API for receipt processing
- Database: SQLite with Drizzle ORM
- Integration: Gmail API for automatic receipt import
Challenges we ran into
- Handling audio file formats across different browsers
- Implementing real-time voice transcription with proper error handling
Accomplishments that we're proud of
- Built a fully functional voice-to-AI pipeline in 24 hours
- Achieved seamless audio transcription with automatic message sending
- Integrated multiple AI services (Whisper, GPT-4, TTS) into one cohesive experience
- Created an intuitive UI that works for both voice and text interactions
What we learned
- Voice interfaces require careful state management and error handling
- OpenAI's APIs are powerful but need proper file handling for audio processing
- User experience is critical - even small delays in voice feedback feel broken
What's next for ExpenseTracker Pro
- Add expense categorization suggestions based on spending patterns
- Implement budget alerts and spending limit notifications
- Expand Gmail integration to support more receipt formats
- Add data visualization for spending trends and insights
- Mobile app with offline voice processing capabilities
Built With
- asr
- oauth
- ocr
- openai
- tts
- typescript
Log in or sign up for Devpost to join the conversation.