Personal project by Praveen Prabhakar.
This app demonstrates how to extract and classify user banking intents from natural language using on-device generative models and the Foundation Models framework. Enter a request such as "Send $50 to Mom tomorrow from checking" and the app will classify the intent and extract relevant details.
- On-device intent classification using Foundation Models
- Supports various banking intents:
- Payments (send money, pay bills)
- Schedule appointments
- Check FICO score
- Find ATMs
- Card management (freeze, unfreeze cards)
- Account queries (balances, transactions)
- Example requests and robust parsing of dates and entities
- Sample input picker and real-time output display
- Open the project in Xcode 26.0 or later.
- Ensure your development environment supports SwiftUI and Foundation Models framework.
- Build and run on a supported simulator or device.
- Enter sample banking requests in the input field, or tap to select from bundled examples.
- View the extracted intent and details in the output box.
Contributions are welcome! Please open an issue or submit a pull request with your improvements.
This project is licensed under your personal license. See LICENSE.txt for details.
This project originally demonstrated the Foundation Models framework, now customized and maintained as a personal project.