AI DISCLOSURE

AI Disclosure Statement

Transparency is core to HargaHantu. Here is a complete disclosure of all AI assistance used in building and operating this application.

AI in Development

This application was built with significant AI assistance using Claude Code (Claude Opus 4.6) by Anthropic. The AI assisted with:

  • Architecture design and technical planning
  • Code generation for React components, API routes, and database schemas
  • TypeScript type definitions and Zod validation schemas
  • Tailwind CSS styling and animation implementation
  • Supabase database setup including RLS policies and RPC functions
  • Performance optimization strategies
  • Accessibility implementation (ARIA attributes, keyboard navigation)

All generated code was reviewed, tested, and refined by the solo developer. The AI served as a pair programmer, not as an autonomous builder.

AI in Operation

HargaHantu uses AI vision models to extract data from receipt images:

PrimaryOpenAI GPT-4o — Vision-based receipt parsing. Extracts merchant name, items, prices, and totals from receipt images.
OptionalAnthropic Claude — Alternative extraction provider. Configurable via environment variable.

The AI extraction is a preview step. Users always review and confirm extracted data before it is submitted to the community database. The AI does not make autonomous decisions about pricing data.

Data & Privacy

  • Receipt images are sent to the AI provider for extraction only. They are not used for AI training.
  • Receipt images are stored privately. Only the submitting user can access their own images.
  • Only anonymized price data (item, price, merchant, area, date) is made public. No user identity is attached to public data.
  • Raw OCR text is not stored. Only structured, validated data is persisted.

Tools Used

Complete technology stack:

Next.js 16TypeScriptTailwind CSS v4SupabaseFramer MotionGSAPThree.jsZodOpenAI APIClaude CodeVercel