InventoryWise AI

The problem

Managing inventory by hand is slow and error-prone. Small and medium businesses struggle to predict the stock they need and lose hours on data entry that creates no value.

The solution

The photo of a delivery note is read by an AI model (Gemini) through a serverless function on Supabase, which keeps the API key safe outside the app. The AI first checks that the document really is a delivery note, then extracts the lines; the user confirms them and the stock updates. A natural language chat answers questions about the warehouse using the actual data. Predictive reordering is still in progress: for now, days of stock are calculated from average consumption.

What it will do

• Delivery note scan: upload a photo, the AI checks it is a delivery note and reads its lines
• Stock updated after your confirmation
• Warehouse chat: «What do I need to reorder?» with answers based on the data
• Days of stock and reorder alerts calculated from average consumption

What it shows

I can integrate artificial intelligence into a real app, using models to solve a concrete business problem and not as a mere special effect.

Next steps

• Predictive reordering
• Real-time sync across devices
• Persistent database on Supabase instead of in-memory data