YoavCodes
@yoav.codes
Technical founder. Building new stuff at my startup lab https://blackboard.sh | 20+ years in early startups | prev: 7 years at webflow You might have met me as Jeff, Joab, or Larry. #startups #ziglang #typescript #bun #electrobun #webflow
My Electrobun 2.0 blog post is already 3400+ words and it's just the intro that covers my thoughts on AI in open source and how the Bun/Zig (Jarred/Andrew) drama affects the future of Electrobun. For those that like long thoughtful reads.
People keep asking me about Cottontail a new JS runtime that'll be available in Electrobun 2.0 alongside Rust, Zig, and Go. Here's a screenshot of how it started vs. how it's going:
I realize I post most my updates to twitter these days since that seems to be the tech center of gravity is these so here’s a quick update on Electrobun.
In the next version of Electrobun 1.18.0: Tons of stability fixes and enhancements. Bumped Bun and Chromium for those that bundleCEF: true. New docs: docs.electrobunny.ai/electrobun/g... Unofficial Raspberry Pi fork led by x.com/francip/stat...
A revolution is coming. Get early access at electrobunny.ai www.youtube.com/watch?v=fTAj...
Electrobunny.ai
YouTube video by Blackboard Technologies
youtube.com
In a few weeks you’re gonna look at “talking to OpenClaw via instant messaging” like it was the stone age. What’s coming are 3KB native apps, written in Typescript (bundle any binaries you want), and rich custom interfaces from anywhere in the world for your entire fleet. electrobunny.ai 🚀
Bunny Ears Is Coming
Bunny Ears is coming.
electrobunny.ai
A month ago I shipped Electrobun v1. A week ago I shipped a deep WGPU integration. A day ago I added 35 desktop apps people are already shipping built with Electrobun to the Readme. An hour ago Electrobun passed 10,000 Github stars.
It's here. Electrobun is ready for you. npx electrobun init blackboard.sh/blog/electro...
Did a pass on Github and found a **ton** of _wildly_ cool projects being built with Electrobun. Added to the readme, you all should check them out.
In the next version of Electrobun (1.16.0): - Merged 18 PRs from the community❤️ - New apis like page zoom & MacOS workspace visibility - Lots of fixes and stability improvements - Kitchen Sink (Electrobun's integration test app) has persistent test filtering and 132 tests & playgrounds - Lot's more
Bunny Ears is coming. Be first in line at electrobunny.ai
Bunny Ears Is Coming
Bunny Ears is coming.
electrobunny.ai
oh if it's just a "quick" scan of my face then it's fine. as long as it's quick. my, and most people's, primary issue with scanning of the face is how long it usually takes. scan away, just don't take too long.
I’m not saying Bluesky is over, but I had 14.5k followers on here who followed me when Electrobun was on hacker news front page for a week straight a year ago, and I had 800 followers on a fairly new twitter account yesterday 24 hours after Electrobun WGPU tweet Twitter: 400 likes Bluesky: 9 likes
It's the WGPU Final Countdown 🚀 👀👀👀👀 www.youtube.com/watch?v=9u5i...
A magician named Gob
YouTube video by Hatsu11
youtube.com
🔪🙈 absolutely **butchered** my name but happy to see wgpu featured on Typescript.fm Will ship it this week and usher in a new era of tiny, powerful games and Typescript-based ML built with Electrobun.
Today on the pod: State of JS 2025 insights, Node's new release schedule, Bun Stage 3 decorators, WebMCP standardization, and community highlights including typesafe Map.has() and Electrobun WebGPU rendering. share.transistor.fm/s/ccaefe60 #typescript
DOOM is nice, but what if you could use a <electrobun-wgpu> html element right next to your <electrobun-webview> elements and instead of a fully isolated OOPIF you get a super powered GPU surface integrated into the host webview's DOM powered by (bun FFI or native FFI) -> wgpu.
Replacing Python as the glue for direct to GPU ML tasks is nice, but what about DOOM 2 ways: - A C implementation (based on DoomGeneric) rendered into a GPU window. - A TypeScript port of DOOM that renders into the same kind of GPU window, using Electrobun’s WGPU APIs from Bun.
So anyway I just came out of the shower to watch the Apple Keynote 🍏 only to find my wife pointing at *this* ⌚️ on the table and my dog 🌭Barkus Aurelius on the couch begging for mercy. 🙏 I then found out there is no Keynote just a bunch of ads on their youtube channel 😭
Replacing Python as the glue for direct to GPU ML tasks is nice, but what about DOOM 2 ways: - A C implementation (based on DoomGeneric) rendered into a GPU window. - A TypeScript port of DOOM that renders into the same kind of GPU window, using Electrobun’s WGPU APIs from Bun.
In a few days you no longer need Python, pip, venv, virtualenv, conda, mamba, poetry, pipenv, uv, pyenv-virtualenv, asdf, system python, homebrew python, conda python. Just native on-GPU ML/AI glue and workloads directly from Bun typescript. No webview, in Electrobun.
While sick in bed, I’ve been playing with blackboard.sh/electrobun/d... and claude code. Several feverish (literally) hours later for the first time ever I have a decent contacts app on my Linux laptop. Next: calendar, notes, email
Electrobun Documentation - Build ultra fast, tiny, cross-platform desktop apps
Complete documentation for Electrobun - build ultra fast, tiny, cross-platform desktop applications with TypeScript.
blackboard.sh
In a few days you no longer need Python, pip, venv, virtualenv, conda, mamba, poetry, pipenv, uv, pyenv-virtualenv, asdf, system python, homebrew python, conda python. Just native on-GPU ML/AI glue and workloads directly from Bun typescript. No webview, in Electrobun.
Games are nice, but Electrobuns new bun -> wgpu unlocks raw gpu compute as well. - lines of code: html: 72, css: 195, browser ts: 318, bun ts: 450. - the ui is using a system webview - the training and inference is done on gpu direct from bun typescript. - dmg is 20.9MB
Games are nice, but Electrobuns new bun -> wgpu unlocks raw gpu compute as well. - lines of code: html: 72, css: 195, browser ts: 318, bun ts: 450. - the ui is using a system webview - the training and inference is done on gpu direct from bun typescript. - dmg is 20.9MB
This is an Electrobun video game app i vibe coded - 800 lines of typescript in bun - 1 bunny logo, carrots are generated - uses Electrobun's built-in Babylon adapter * wgpu window - no webviews - dmg is 21MB (basically the bun runtime + wgpu compressed) - updates as small as 4KB
This is an Electrobun video game app i vibe coded - 800 lines of typescript in bun - 1 bunny logo, carrots are generated - uses Electrobun's built-in Babylon adapter * wgpu window - no webviews - dmg is 21MB (basically the bun runtime + wgpu compressed) - updates as small as 4KB
three.js is nice, but what about a robust adapter for working with Babylon’s WebGPUEngine directly from bun without needing a webview.
three.js is nice, but what about a robust adapter for working with Babylon’s WebGPUEngine directly from bun without needing a webview.
fam what part of don't retweet this don't you people understand?
fam what part of don't retweet this don't you people understand?
top secret fam. DO NOT SHARE EXTERNALLY. the little window that opens up from the Electrobun Kitchen Sink *does not contain a webview*. - no dom - no cssom - just a native window - with direct access to webgpu/metal - three.js api layered on top - directly in bun typescript
top secret fam. DO NOT SHARE EXTERNALLY. the little window that opens up from the Electrobun Kitchen Sink *does not contain a webview*. - no dom - no cssom - just a native window - with direct access to webgpu/metal - three.js api layered on top - directly in bun typescript