Life Screenshot Organizer – Project Story

Inspiration

We noticed that people take screenshots constantly — whether for shopping, studying, or saving important information — yet most screenshots get forgotten or never acted upon.

We asked ourselves:

"What if screenshots could tell us why they were taken and suggest actionable steps?"

This inspired Life Screenshot Organizer, a system that turns passive screenshots into actionable insights, helping users reduce digital clutter and act on what they save.

What it does

Life Screenshot Organizer is an AI-powered system that:

1. Detects Intent – Each screenshot is assigned an intent:
   
   • buy → for shopping/comparison
     
   • study → lecture slides, notes
     
   • remember → quotes, events
     

Gemini 3 Flash Preview outputs a confidence score (C \in [0,1]):

$$ I, C = \text{Gemini3}(\text{screenshot}) $$

1. Groups Related Screenshots – Screenshots with similar categories and intents are clustered together. The cluster size is:
   

$$ |Cluster_k| = \sum_{i=1}^{n} \mathbf{1}(category_i = k) $$

1. Suggests One Smart Action – For example:
   
   • "You saved 5 laptops. Want a comparison table?"
     
   • "These study slides are related. Generate revision notes?"
     

All reasoning, clustering, and action suggestion is handled entirely by Gemini 3 Flash Preview, without OCR or extra comparison logic.

How We Built It

• Frontend: HTML + CSS + Vanilla JavaScript for a fast, clean MVP.
  
• Backend: Node.js + Express for file uploads and metadata storage.
  
• AI Engine: Gemini 3 Flash Preview for multimodal reasoning.
  
• Storage: Local JSON files store screenshot metadata:

{
  "image_name": "screenshot1.png",
  "intent": "buy",
  "category": "laptop",
  "cluster_id": 1,
  "action_status": "pending"
}

Built With

• epress
• gemini3
• node.js
• react
• tailwindcss