This is a submission for the MLH x DEV Writing Challenge
What I Built
I built Vision Assistant v1.0.0, a custom desktop AI vision tool created from scratch using Electron.
-
What it does: It brings instant, system-wide AI assistance directly to your desktop environment without the friction of constant context-switching. It features global shortcut activation (
Ctrl + Shift + Space), multi-screen capture capabilities, real-time chat & vision analysis, and memory persistence across sessions. - The Inspiration: I wanted to eliminate the workflow friction of switching between browser tabs and separate tools when analyzing multiple screen areas or seeking quick AI assistance during development.
-
What I Learned: Building a desktop app from scratch using Electron taught me a lot about handling system-level global shortcuts, managing multi-screen capture streams, and packaging a clean Windows executable using
electron-builder.
Demo
You can check out the setup guide video and walkthrough here:
Vision Assistant Setup Guide Video
You can also download the latest Windows standalone release package directly from the GitHub repository:
GitHub Repository & Releases
release link for windows user
Partner Technologies
For this project, I focused heavily on core web and desktop ecosystems using JavaScript (69%) and HTML (31%), powered by Electron to bundle a seamless standalone Windows application with a dark/purple customized UI theme optimized for desktop environments.
Hackathon Experience
Building Vision Assistant for this challenge was an incredible journey of pushing desktop development boundaries. Bringing an idea from local scripts to a fully packaged, global shortcut-enabled desktop application made it a deeply rewarding learning experience!
Top comments (0)