UI, UX, 3D. Cooking something new. Prev: @LumaLabsAI @theatre_js @splinetool
🌎
Joined November 2015
- Tweets2.5K
- Following2.1K
- Followers5.4K
- Likes7.3K
Pinned Tweet
Dear frontend devs and UI designers. I bring you Liquid DOM, a complete and faithful implementation of Liquid Glass on the Web.
- Shape morphing
- All properties animatable
- Dynamic refraction and reflection
- Adaptive tint
- Adaptive specular highlight
- Dispersion
- Full html integration
- Super fast layout engine that works across Canvas and html
- Pointer event handling
- Framework and renderer-agnostic low level API
- High level React API
- Ootb @threejs and r3f integration
And lots more.
Read on for implementation details and demos.
Andrew Prifer (is hiring) retweeted
Some random things I need to say out loud:
1) AI so far is dangerously converging power in the hands of few and increasingly swapping labor for capital/compute
2) In a hyper capitalist environment that’s a recipe for disaster without rethinking income and what gives life purpose
3) Some of you are way too eager to give up on your own thinking
4) Some people and ideas are a cancer to democracy and society. Support is a choice. So is not associating.
5) We are at an all time low for ethics, principles and morality
6) Even thinking of AI welfare before welfare of all humans is deeply immoral
7) New Siri still sucks
8) I don’t quite get the iPhone Duo
9) The Mission S is beautiful
Andrew Prifer (is hiring) retweeted
My dear friends, I’d like to show you my latest toy project
Introducing Handwired: turn your regular JavaScript code into a neural net!
- no training
- exact outputs vs the JS program
- fully differentiable
All runnable in a browser, without GPU
we now live in an age when maintaining your own JS interpreter is considered less complex than using node.js isolates
from the @opencode 2.0 blog
“ugly solutions” in math have been making the rounds on X lately, and the fact that the new square packing record looks like shades and the decimals start with .69 is a redemption arc if I’ve ever seen one
I find it so ironic that the 1.85:1 aspect ratio was invented in the ‘50s as a way for cinemas to compete with the then predominantly 4:3 TV screens by offering an exclusive format, and today the exact same thing is happening, just in reverse.
In IMAX theaters, ‘Godzilla Minus Zero’ will expand to a floor-to-ceiling aspect ratio for the final 75 minutes of the film.
(source: broadwayworld.com/bwwtv/arti…)
Andrew Prifer (is hiring) retweeted
Experts doubted a small neural net could beat Craftax without an LLM's knowledge.
Our agent meta-harness built one in 3 weeks. The 15M-param agent beats the final boss 44% of the time, costing a cent, and 10 seconds.
GPT-6 Astra won 1%, taking 24 hours and $83. 🧵
The silver lining of one of the year's best movies flopping so hard is you can get a private screening at your local imax for 20 bucks
I wanna say his memory lives on, but that would be technically incorrect
OpenAI, @thsottiaux, fellas, this is why you can’t anthropomorphize these things and give them cute avatars and names. Because are you telling me I’m gonna have to now euthanize my little froggo because its context is eternally poisoned from that stupid refusal earlier?
Welcome Trish, may you live longer than Froggo (RIP 2026-2026)
OpenAI, @thsottiaux, fellas, this is why you can’t anthropomorphize these things and give them cute avatars and names. Because are you telling me I’m gonna have to now euthanize my little froggo because its context is eternally poisoned from that stupid refusal earlier?
OpenAI, @thsottiaux, fellas, this is why you can’t anthropomorphize these things and give them cute avatars and names. Because are you telling me I’m gonna have to now euthanize my little froggo because its context is eternally poisoned from that stupid refusal earlier?
I fear I’ve gotta stick with Grok Bot for now. Giving in to demands like Amazon’s is a slippery slope to most useful things a regular person might want to delegate being siloed away from personal AI assistants.
Andrew Prifer (is hiring) retweeted
More like same entry point, same capabilities, shared context and shared organization features. And sure, you can adapt the UI for the specific task, like professional engineering work.
The current lineup looks like a collection of legacy versions of the same product they tried very hard to mold into separate offerings for different user personas instead of deprecating them, but imo it's just confusing and backs them into a corner.
Local codex is the one bit that's tough to integrate as something that is inherently malleable and used as a local tool, and that's where it would instantly help to demarcate not around the user persona (codex for engineers, work for office workers), but around whether the conversation, memory, context and available capabilities are local only or available in the cloud. That then also results in a surface they can iterate and improve on much more easily because the distinction is based on real constraints and not on theoretical user personas, and as a result it is also much clearer for the user what they can and cannot do.
OpenAI I’m saying this with all the love of someone who’s already upgraded to that comical $500 tier, what are we even doing here man.
- ChatGPT: can execute agentic tasks, run code, create documents, interact with apps
- ChatGPT Work: can execute agentic tasks, run code, create documents, interact with apps, but has a VM and a browser
- Codex: can execute agentic tasks, run code, create documents, interact with apps, has a VM and a browser, but shows you file diffs
- Dots: can execute agentic tasks, run code, create documents, interact with apps, has a VM and a browser, but has a cute mascot
Framing great feedback as a dunk is a bit counter-productive imo as demonstrated by all the comments either further dunking or being defensive.
Because this is otherwise the kind of feedback you do want to see. Detailed, specific, based on personal pain points, not hypothetical.
I am once again trying Paper and once again I will tell you this is truly an example of how not to do onboarding. I mean this as constructive criticism, but hopefully everyone can learn a thing or two as this is an important investment you can make in your own products and it is _entirely in your control_.
1) for starters, there's basically no actual onboarding
2) there's a small widget to setup MCP, which I hear you basically must use to get value out of it
3) you click it, you have to install a desktop app - ok fine odd, its 2026, but whatever
4) there's then once again no onboarding, its just the web view wrapped in a desktop app
5) click a new tiny widget in the corner and get sent to docs
6) docs tell you to click a button that doesnt exist
7) be willing enough at this point to continue, create a new document or click into one, then you'll find a buried connect button (or the same button that tricked you previously) and you can finally register the MCP
there are so many ways you could improve upon this, but i really hope folks take this seriously. it doesnt matter how good your product is if you dont make it easy for new customers to use it, and thats more true now than ever with the market expanding
sorry this is ranty but i really hope people consider this stuff