This is a submission for the Gemma 4 Challenge: Build with Gemma 4
What I Built
Vellum is a tray‑resident macOS app that turns any screenshot into a conversation. Hit ⌘⇧1 to drag a region, or ⌘⇧2 to grab the whole screen — the capture pops open in a chat window, the image is described and OCR'd in the background, and you can immediately ask follow‑up questions about what's on screen.
The interesting bit: Vellum runs Gemma 4 locally on Apple Silicon by default. No screenshot ever leaves the machine. There's no account, no API key, no cloud round‑trip — just a global hotkey and a vision‑language model living in your menu bar.
How I actually use it: over a few days I capture screenshots of things I read on the internet — articles, threads, diagrams, half‑finished thoughts — and let them pile up. Then one evening I sit down and review the whole stack. Because every capture is OCR'd by Gemma 4, the pile is fully searchable, and the review session turns into something more useful than a screenshot graveyard:
