GetZippo

GetZippo is a zero-friction, privacy-first utility that uses Gemini’s multimodal to architect, categorize, and ZIP your files in seconds.

Comment
  1. Inspiration (The "Why")

We’ve all been there: a desktop littered with Screenshot_2026_02_15.png, Final_Final_v2.pdf, and random receipts. Manually sorting these is a chore, and most AI tools want to own your data. We wanted to build a "Digital Janitor"—an AI that cleans your room without looking at your secrets. GetZippo was born from the need for a keyboard-driven, privacy-first utility that turns a mess into a structured .zip in seconds.

  1. What it does

GetZippo is an AI-powered file architect.

Drop & Go: Drag a mess of files into the "Void."

Multimodal Analysis: It uses Gemini 2.0 Flash to "see" images and "read" PDFs.

Semantic Organization: It doesn't just sort by file type; it sorts by context (e.g., "University/SDA/Quizzes" or "Finance/Receipts/Kuwait_Trip").

Smart Renaming: It gives every file a human-readable name based on its actual content.

Instant ZIP: One click (or keystroke) to download your newly structured life.

  1. How we built it

    Frontend: A sleek, minimal Next.js application styled with Tailwind CSS.

    AI Core: Integrated Gemini 2.0 Flash via OpenRouter for high-speed, multimodal inference.

    Processing Pipeline: * Client-Side Pre-processing: We implemented local image compression (max 1024px) and PDF extraction (first 2 pages) to minimize latency and token costs.

    The "Architect" Prompt: A complex system prompt that forces the LLM to output a valid JSON directory structure.

    Deployment: Hosted on Vercel for 100% uptime and fast global delivery.

  2. Challenges we ran into

    Context Consistency: Ensuring the AI remembers the folder structure it created for the first 5 files when it processes the next 5. We solved this by using Multimodal Batching.

    Latency vs. Accuracy: Balancing the speed of Gemini 2.0 Flash with the need for deep document understanding. We optimized this by using a Hybrid OCR-First approach.

    Mouse-less UX: Designing a web interface that feels like a Linux terminal—efficient, fast, and completely controllable via keyboard.

  3. Accomplishments that we're proud of

    Privacy by Design: Building a tool that handles sensitive data without a persistent backend database.

    The "Magic" Moment: The first time we dropped 20 random files and saw them perfectly categorized into a nested folder structure in under 10 seconds.

    Zero Friction: Creating a utility that requires zero signup. You land, you drop, you zip.

  4. What's next for GetZippo

    Deep OS Integration: Moving beyond the browser to a CLI tool for power users.

    Local LLM Support: Optional integration for running models like Llama 3 locally for 100% offline privacy.

    Custom Architectures: Allowing users to provide a "Template" (e.g., "Organize this like a FAST-NUCES semester folder").

Built With

Share this project:

Updates