JZ

Creating next-gen creative software for both humans and AI @StewdioAI

[JZ @jz2letters]

@Tezenari @shadcn I tried switching to zed many times but its auto-complete seems slower and less intelligent than cursor. @zeddotdev can we get it fixed please? I really want to love it

[JZ @jz2letters]

Yo guys how do you make money from open source? Want to open source a creative tool but I also wanna make money (we live in a capitalist society after all)

[Blake Shao @blaketshao]

sumin crazy is coming from @jz2letters

animate in claude code and have llm expose bespoke controls, this hit a spot

[JZ @jz2letters]

@steveruizok Have you tried @gitbutler ? I think it's exactly made for this use case

[JZ @jz2letters]

@NathanFlurry I feel like @jhleath would have a lot to say about this 🤔

[JZ @jz2letters]

There's a subtle difference between code as data vs code as the de facto renderable output. @paper seems to be the former, similar to existing site builders like @framer, meaning that instead of storing the raw HTML that the browser renders, they store it as some sort of serializable structured format like JSON.

This is important because the 2 mental models differ greatly on expressiveness, authoring model, and human editability.

Expressiveness: HTML is ultimately expressive since that's literally what the browser renders. Almost anything that can be rendered by the browser can be expressed in HTML. The JSON to HTML approach has to constantly be playing the catch-up game because they basically need to re-invent HTML in their format. Not sure if it can truly be feature complete.

Authorship: Besides, LLMs have seen tons of raw HTML + CSS examples during training but none of this new format. This means that AI's performance would be compromised. Unless you're goated like @QuiverAI and train a model on your format (but you're still actively iterating on the format, so not sure if it'll work either). Alternatively, you let AI write HTML and convert it to your format, then you run into the first problem of expressiveness.

Editability: Good thing is, storing structured data and rendering to HTML is way more editable. You control the format, so you can build the right UI for humans to interact with the data. Otherwise, you'd be writing some sort of HTML patching tool.

Great read!

[Yoko @stuffyokodraws]

https://t.co/K5FtFSAFKt

[JZ @jz2letters]

@blaketshao @greensock This looks sick! Can we add this to our product?

[JZ @jz2letters]

I was trying so hard to make worktrees work for my workflow but ended up with @gitbutler and it's SOOO much better

[Theo - t3.gg @theo]

3. Worktrees are an abomination

We need source control that makes more assumptions about our file systems. Why do we have to implement copy on write ourselves? Why can't I "check out" the same branch in 2 places at the same time? Why do I have to put so much energy in keeping my worktrees updated to main?

We've been taken for fools. The current worktree implementation in git should be ignored entirely. Our agents deserve better.

[JZ @jz2letters]

New minecraft just dropped

[Blake Shao @blaketshao]

brutalist structure exploration with @threejs, with light cube marching application inspired by @OskSta's super cool townscaper

[JZ @jz2letters]

@blaketshao @studiodumbar @OpenAI @greensock Woah how did you do it?

[Nathan Baschez @nbaschez]

Would be cool if there was an easy way to share HTML pages

You could imagine a world wide web of these documents, linked together

Really cool idea

[JZ @jz2letters]

@konstipaulus 🐐

[JZ @jz2letters]

@albert_ Huge if true

[JZ @jz2letters]

@vimtor @PlanetScale Is SST still being actively developed? Claude told me y'all slowed down development to focus on OpenCode

[Wes Bos @wesbos]

Me: why would anyone watch sports, what a waste of time!

Also me: ohhh what did the one hosting company employee say to the other hosting company employe today?

[Trata (YC W25) @trytrata]

Trata (@trytrata) is now on FinTwit. We host anonymous buyside to buyside conversations via voice calls. We work with 125+ funds (vs. 10 in March) and receive 25-30 human-to-human calls weekly. We compile these conversations into a library of investment research.

EX: Fund 1 is bullish and Fund 2 is bearish on $OPEN, they ask each other questions they want answered surrounding the key debates for anywhere between 30-60 minutes.

The average fund on the platform: ~$200-400mm in AUM, 3 investment professionals, fundamental research focused. These funds use tools like Bloomberg and third-party expert networks, and have regular access to management teams.

To the qualified lurkers and anon accounts of FinTwit: Request a call on any ticker (bull or bear). If we pair you, we’ll give you:
1. Access to that transcript and any transcripts on that ticker published through the end of 2026
2. Airpods

[JZ @jz2letters]

@artwormm @APazyryk @AaronRegunberg Did you see the press release?? The head of police quite literally said that it was the FBI who followed that lead and was the reason why Benjamin was detained. And yes - we do have a problem with Benjamin's information doxxed. It's absolutely unacceptable!

[Y Combinator @ycombinator]

Freeport (@Freeportmkts) builds intelligent, research-backed crypto portfolios — integrating the best of DeFi to earn real yield, diversify across trusted assets, and stay in self-custody.

No fees, no noise. One page, one click.

https://t.co/XOFhsziMX0

Congrats on the launch, @Lihong062 & @brybryman!

[JZ @jz2letters]

@LandseerEnga @StewdioAI 😭bro call me let me onboard you

[JZ @jz2letters]

@rmptmlk + all the generative models on Stewdio (YC F24) killed it again 😎

[Y Combinator @ycombinator]

🎨@StewdioAI gives creatives real control over AI-generated visuals.

Creatives are already using Stewdio for music video + visualizers (including work for Quavo and Loco), advertising storyboards, and architectural renderings.

Congrats on the launch @jz2letters & @rmptmlk!

https://t.co/G3OEeUodYL