Page 609 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

Simple solutions:

• Don’t store full card numbers (only last 4 digits if needed).
• Use strong encryption (AES-256 is already used — great!).
• Enable Multi-Factor Authentication (already included — well done). 4.6: Finance

3) What types of models are needed? (OCR, classification, recommendations, etc.)

Here are useful models:

Model Type Purpose
OCR Read text from receipts (already done)

Image Processing Improve photo quality (fix blur, angle)

Classification Categorize receipts (e.g., food, travel)
Recommendation Suggest budgets or savings goals

Prediction Forecast future spending

4) What is the current status of the project?

The application is largely working, but it needs some modifications. It is currently in the actual
testing phase and is being developed.

• Internal testing (UAT) completed.
• Feedback is being collected to improve the app before public release.

5) Can the app connect with WhatsApp, WeChat, or Email?

Yes, and it’s a very useful idea.

Simple solutions:
• Email: Auto-scan receipts from emails (e.g., Gmail).
• WhatsApp: Use WhatsApp Business API to import receipts.

5 References

[1] Smith, A. (2022). Optical Character Recognition: Best Practices for Receipt Digitization.
TechJournal

[2] Deploying Mobile Apps with AWS Amplify. AWS

[3] Mobile Application Security Verification Standard (MASVS). OWASP

[4] Nielsen, J. (2021). Usability Testing Guidelines. NNG

[5] Automating Deployment with GitHub Actions. GitHub Docs.

573