1. Information Collection Architecture
Hisaabak processes operational metadata explicitly required to power your digital storefront:
- •Authentication Data: Email addresses and user identification tokens managed securely through Google or Supabase Auth workflows.
- •Inventory Images: Product photos captured during the SOSAD workflow or scanner sessions are processed exclusively for barcode and visual item cataloging.
- •Ledger Metrics: Sales items, stock quantities, and pricing structures required to construct your dashboard analytics feeds.
2. AI Engine Proxy Processing
To minimize latency and optimize retail point-of-sale speeds, item photos are routed through a secure Next.js server proxy middleware.
This proxy isolates third-party API keys and dispatches temporary, anonymous image payloads to the Gemini 1.5 Flash API for instantaneous label suggestions.
Image processing data blocks transmitted to the AI core are utilized solely for contextual identification and are never stored or retained for model training pipelines.
3. Multi-Tenant Row Isolation
Your store ledger analytics, custom pricing metrics, and registration credentials are fully compartmentalized. All database pathways enforce structural PostgreSQL Row Level Security (RLS) layers, restricting access strictly to authenticated shop operators tied directly to your unique identifier.
4. Browser & Native API Permissions
Because Hisaabak functions explicitly as a hardware-free POS replacement, your web application will request the following access tokens:
- •Camera Interface: Utilized strictly for executing barcode or camera sight scanning actions.
- •Haptic Vibrations: Calls the device Web Vibration API to deliver tactile scan feedback flags.