Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

I Built a Roman Urdu Stock Book for My Friend's Grocery Store

Description and use

Stock Diary turns mixed-language stock notes into reviewable ledger entries, providing real-time balances and low-stock alerts for a small grocery store.

Problem

Hamza’s grocery store stock was kept in paper and memory, mixing Urdu, Roman Urdu, and English, making accurate tracking difficult.

Outcome

Stock Diary converts mixed-language stock notes into accurate, reviewable ledger entries, enabling real-time balance tracking and low-stock alerts for a small grocery store.

Workflow

  • 1. User types a stock movement sentence (e.g., "20 carton basmati aaye, 5 packet haldi nikle").
  • 2. Gemma 4 parses the sentence into proposed movements with product, quantity, unit, direction, and balances.
  • 3. The app displays the proposals and highlights any unit or quantity errors.
  • 4. User reviews and confirms the movements; only confirmed entries are written to the ledger.
  • 5. The ledger updates balances and flags low-stock items.
  • 6. User can ask natural language queries (e.g., "kis cheez ka stock kam hai?") and receive answers based on the ledger.

Evidence

Stock Diary turns “20 carton basmati aaye, 5 packet haldi nikle” into two proposed stock movements.

Source

The model helps read the sentence; plain Python checks the numbers and posts the confirmed movement to a SQLite ledger.

Source

Gemma only works out which question he is asking; Python answers from the ledger.

Source

  • Record date: 2026-10-04T17:55:57.190Z