Post · Agents & browsers
Computer use without screenshots
On-device segmentation and OCR turn the screen into text for Jev.
Open x.com ↗Okay so Jev can actually do computer use really well
Without any screenshots, or LLMs and no Pixels leave my mac
I dont even read the Dom elements
A local CoreML model segments every button and UI element on screen.
On-device OCR reads the labels. That text is all Jev gets.
How builders describe it
Computer use with Jev and local vision model! Its so goddamn fast
The decision Jev makes
Pick the next click, field or navigation from the legal actions on screen.
Where it fits
Browser agents need to choose an action on every step. With Jev each step is a choice from the actions actually available on the page, which keeps agents fast and on the rails. All 299 agents & browsers projects →
Related projects
Agents & browsersdocent_chrome_pluginA pluign for TypeSafe AI's Jev.GitHub · ★ 2 · yepher
Agents & browsers@sideform/browser-agentGoal-driven browser agent library in TypeScript (controller, step policy, browse_task loop) whose step policy uses a System One decision client such…GitHub · dan-rozhkov
Agents & browsersLathAgent loops that pay for model calls over a Tempo payment channel with every call recorded on Solana; a separate TypeSafe Jev decision answers…GitHub · lath-agents
Agents & browsersvoice-browsermacOS command bar that drives Chromium, Dia, Arc or Chrome from plain English; TypeSafe Jev (or SubQ Snap) picks clicks and targets for fast single…GitHub · ken-code-wq
Agents & browsersjevdohsom (Harness for System One Models, repository jevdo) lets Jev operate a bash shell by walking a strict-allowlist decision tree of commands,…GitHub · ★ 4 · gi-dellav
Agents & browserskureksistantAutonomous, sub-second personal AI assistant for Arch Linux (Hyprland) & macOS.GitHub · ★ 1 · nodaysidle