Sortify

Overview

Sortify is a document classifier website. Upload untitled documents and an LLM (the Gemini API, or a local model through Ollama) fully scans each one and generates an accurate title.

A checklist mode goes a step further: list the documents you need, and it identifies which files contain that content.

The problem

Scans and downloads pile up as scan_0012.pdf and IMG_4417.jpg, and finding the one document you actually need means opening them one by one.

How it works

• Choose your model. Titles come from the Gemini API, or from a local model through Ollama.
• OCR first. Images and large files go through OCR, so the model receives text instead of raw pages and token usage stays down.
• Checklist matching. Enter the documents you expect, and Sortify identifies the files that contain each one.
• Organized export. Download all the classified material as an organized ZIP folder.

Key features

• 01

  AI titling

  Accurate titles generated from each document’s content.

• 02

  Cloud or local

  Gemini API or Ollama.

• 03

  OCR pre-pass

  Keeps token usage down on images and large files.

• 04

  Document checklist

  Find which files contain what you need.

• 05

  ZIP export

  Everything classified, organized and downloadable.

• 06

  Drag & drop

  Files or whole folders: PDF, JPG and PNG.

Challenges & what I learned

Token budgets shape the whole design. Sending every page of every file to a model is slow and expensive, which is why OCR does the heavy lifting before the LLM ever sees a document.

What’s next

A public prototype link, and smarter folder structures inside the exported ZIP.