FridgeSloth 🦥: The 100% Local Fridge-to-Feast Copilot & Grocery Butler Powered by Gemma 3

Description and use

FridgeSloth is a local, on-device AI that scans fridge contents with Gemma 3 Vision and suggests minimal‑effort recipes, all while keeping data private and free of cloud costs.

Problem

People spend 14 minutes daily staring at an open fridge, experiencing decision paralysis and wasting energy.

Outcome

FridgeSloth reduces decision paralysis to a 1‑minute photo capture, delivers a 99% effortless recipe in under 4 minutes, and generates a grocery list sorted by aisle—all while keeping data local and free of cloud costs.

Workflow

• 1. Capture a photo of fridge/freezer/pantry.
• 2. Send image to local Ollama server running Gemma 3 Vision.
• 3. Model returns ingredient list and recipe suggestions.
• 4. App displays recipe steps, prep/cook times, dishes to wash, and a grocery restock list.
• 5. User follows minimal-motion recipe and optionally restocks items.

Evidence

The average person spends 14 minutes every day standing motionless in front of an open refrigerator with the door ajar

Source

Gemma 3 (4B Multimodal Vision) runs comfortably on laptop hardware (under 3.4 GB of RAM/VRAM).

Source

Zero cloud uploads. Photos of the inside of your home never leave your device.

Source