Wednesday, October 7, 2026ArchiveSearchAsk the paper

The Computomatix Times

All the posts fit to save — curated from @computomatix's bookmarks & likes on X

Claude Code Paired With Google Stitch 2.0 Redesigns Vibe-Coded Apps

Claude Code + NEW Stitch 2.0 just changed how I design apps

Prajwal Tomar argues that generic AI-built app UIs are a workflow problem and describes using Google Stitch 2.0 with Claude Code via MCP to generate a design system and redesign a client app in under an hour.

Original post · 11 min read
X ArticleClaude Code + NEW Stitch 2.0 just changed how I design apps
Google Stitch 2.0 + Claude Code via MCP is the workflow I’ve been testing… and the results are genuinely insane.
I've been saying this for a while now. If your app looks like AI slop, that's not an AI problem. That's a workflow problem.
Most builders are stuck in the same cycle. You build something functional with AI. The features work. The logic is solid. But the UI looks generic and you know it. Your users know it too. And the moment someone lands on your product, they make a judgment in about two seconds.
The old fix was to hire a designer. Spend $3,000 to $10,000. Wait weeks for Figma files. Then spend even more time getting someone to actually implement those designs into your codebase. By the time the design was live, you'd already lost momentum and burned through cash you didn't need to spend.
That entire process is now optional.
I was able to take a client project that looked like every other vibe coded app and completely redesign it in under an hour. Professional typography. Consistent color system. A design language that actually holds together across every screen.
This is the workflow I'm running now. And I think every builder shipping with AI needs to understand how it works.
Why Stitch 2.0 Is Actually Different
I've tested a lot of AI design tools over the past year. Most of them generate something that looks decent on the first screen and then falls apart the moment you try to build a second page. Consistency is the problem. It always has been.
Stitch 2.0 solves this in a way nothing else has.
It's an AI native canvas. You can feed it screenshots of your existing app, drop in inspiration images from places like Dribbble or 21st.dev, or even paste a URL and let it reverse engineer the design of any website you like. It takes those creative seeds and generates full UI designs with multiple variants you can pick from.
But the real unlock is not the designs themselves. It's what Stitch builds in the background while it's designing.
A complete design system.
Typography scales covering display, headline, label, title, and body fonts. A primary, secondary, and tertiary color palette that's auto generated to be complementary. Color scales for every shade. Component rules and patterns. Elevation and depth specs. Even dos and don'ts for your design language.
All of this gets documented automatically in a file called design.md. This is a plain markdown file that captures every single design rule in one place. And this is the file that changes everything when you bring Claude Code into the picture.
How I Actually Use This on Real Projects
Let me walk you through exactly what I do. No theory. Just the workflow.
I start by taking screenshots of whatever I'm redesigning. If it's a client project or something I've been building myself, I screenshot the main screens and drop them directly onto the Stitch canvas. If I'm starting fresh, I'll grab two or three inspiration images from Dribbble instead. You don't need more than that. You're not looking for something to copy. You're looking for a direction.
Then I write one focused prompt. I tell Stitch what the app does, which screens I want redesigned, and the design direction I'm going for. Dark mode, minimal, editorial, whatever fits the product. I also specify font preferences because fonts are honestly the single fastest way to elevate how an app feels. Serif for headings, clean sans serif for body is a combination that works really well for most SaaS products.
Stitch generates multiple variants from that one prompt. This is important. Don't just accept the first output. Look at each variant and pull the best elements from each. The typography from one, the layout from another, the color energy from a third. You're curating, not just accepting.
One thing most people miss about why Stitch produces such strong output is that it generates images first before any code. That means it's not constrained by what HTML and CSS can do. It can imagine anything visually. Then you work backwards from that reference to build it. That's why the designs feel so much more polished than what you get from prompting a coding tool directly.
You can also talk to Stitch via voice now instead of typing. It transcribes your words into prompts automatically. Small feature but genuinely useful when you're deep in a session and want to keep moving fast.
Design.md Is the Real Game Changer
Once you're happy with your designs, go to the right hand panel in Stitch and click on Design Systems. You'll see that Stitch has already created one for you automatically based on everything you've been designing.
Click into it and you'll find the full design system I described earlier. Typography, colors, components, rules, everything documented and organized.
Now click on design.md.
Copy the entire file. Go to your project. Create a new file called design.md in the root directory. Paste it in. Save it.
That file is now the single source of truth for your entire design language.
Here's why this matters … continue on X ↗
♥ 1.0K · ⟲ 103 · 👁 805.5KView on X ↗

More in Agents & Dev Tools

Developer Rebuilds Seven Adobe Apps in Rust Using Opus 5.5

Peter Yang highlights a developer who reimplemented seven Adobe apps, including Photoshop, Premiere and Lightroom, in Rust with Claude Opus 5.5 and open-sourced them. The developer believes they can match Adobe's features within months, against Adobe's $840 yearly all-apps plan.

Original post · 1 min read
It's insane to watch AI blow apart closed source software and games.

4 examples from the past month:

1. 7 of Adobe's biggest apps, including Photoshop, Premiere, and Lightroom, have been partially rebuilt in Rust with Opus 5.5 and open sourced. It's still early, but the developer thinks they can match Adobe's features within months. Adobe's all-apps plan costs $840/year.
Miguel Ángel Durán @midudev
Todos los productos de Adobe reimplementados desde cero, gratuitos y de código abierto

→ getartcraft.com/apps
♥ 50 · ⟲ 2 · 👁 10.7KView on X ↗

Vercel's Guillermo Rauch Explains Turborepo's Migration From Go to Rust

Guillermo Rauch says Vercel moved Turborepo from Go to Rust, a migration that was controversial internally due to human costs. He argues that with AI agents the calculus has changed, so what is best for humans is no longer necessarily best for business.

Original post · 1 min read
DHH is fundamentally right about Rust. For context, Vercel has been undergoing a Rust-ification (carcinization, technically 🦀) for a while.

One of the first projects we migrated was Turborepo, from Go to Rust¹. The migration completed, but the RoI was actually quite controversial internally.

While Rust was in our eyes better for low-level OS access, something crucial for a build system like Turbo, the human migration costs were very sustantive.

Go is very fast. It's beautifully designed. It's easy to iterate on. We were very conflicted about the migration, because it was *humans* writing the code, *even if we knew Rust was a better choice*.

The calculus has now changed. What's "best for humans" is no longer necessarily "best for business".

FWIW, it's also quite unlikely that Rust is the end-all-be-all toolchain. I'm quite certain there's greener pasture ahead, because Rust itself was designed before the 'supersonic tsunami' of agents hit.

¹ https​://vercel.com/blog/how-turborepo-is-porting-from-go-to-rust
♥ 3.2K · ⟲ 152 · 👁 352.7KView on X ↗

Integer Multiplication Algorithm Bound Tightened Repeatedly With Astra

A post reports that a user running ChatGPT Astra in a loop is repeatedly breaking records for integer multiplication algorithms. It quotes an update to OpenAI problem #109 that tightens the constant from 2^-182 to 2^-59, a roughly 500,000-fold improvement over the previous result.

Original post · 1 min read
This guy has 6.1 Astra running in a loop and is breaking the record for integer multiplication algorithms every few hours lmaooooo.
Doug Colkitt @0xdoug
We are publishing an update to OpenAI problem #109 Integer multiplication) with another substantial further tightening:

κ = 2⁻⁵⁹ (from OpenAI’s original κ = 2⁻¹⁸²)

Approximately 500 thousand fold improvement over our previous result and a 2¹²³ fold improvement over the original OAI result.

The latest redesigned the finite network to share intermediate computations and scratch space, then tightened the recursion and Gaussian estimates.
♥ 4.1K · ⟲ 118 · 👁 167.4KView on X ↗

Boris Cherny Says Prompting Claude Should Feel Like Talking to a Coworker

Boris Cherny explains his approach to prompting Claude, advising users to give clear goals, specify effort level and verification steps rather than relying on heavy scaffolding.

Original post · 1 min read
I am surprised that people are surprised this is how I prompt Claude.

Talk to Claude the way you would a coworker. There's no secret to prompting. There's no need to be overly scaffolded or prescriptive for most tasks -- give Claude a goal, and it will figure it out.

Back in the Sonnet 3.5 days, your prompt mattered a lot. Nowadays, it's much more important to communicate to the model:

1. What you want it to do
2. How much effort you want it to spend
3. How it should verify that it did the right thing
Boris Cherny @bcherny
Prompt
♥ 12.1K · ⟲ 720 · 👁 1.1MView on X ↗

Eric Raymond Highlights Open-Source Rust Clone of Photoshop Built via LLM

Eric S. Raymond shares the photocraft GitHub project, a clean-room open-source reimplementation of Photoshop that he says was likely generated by decompiling the app, converting it to a spec and prompting an LLM for Rust. He argues this threatens closed-source software.

Original post · 1 min read
This is the doom I predicted a few days ago, coming for Photoshop. A clean-room open-source reimplementation.

No prizes for guessing that they decompiled Photoshop to source code, processed that to some kind of non-code specification language, then fed the spec to an LLM with an instruction to generate Rust.

Adobe just got nuked. And closed source is dead, dead, dead.

github.com/storytold/photocraft
♥ 16.2K · ⟲ 1.4K · 👁 3.7MView on X ↗

Nat Eliason Details Fourteen Ways His Bot Setup Automates Work

Nat Eliason lists fourteen functions of his bot setup, including a chief-of-staff agent that drafts emails, specialist agents per work lane, and cloud coding agents that open pull requests from Linear issues. He notes GrokBot as a substantial improvement over his previous OpenClaw setup.

Original post · 2 min read
Things my @bot setup does that still blow my mind:

1. A Chief of Staff who opens the day pulling open loops from email & tasks and suggesting things it can knock out before 7am.

2. After every meeting, decisions get folded into Notion, Linear, and Todoist — not left rotting in Granola

3. Every email starts as a draft. The CoS bot scans my email every ~2hr and drafts replies to nearly everything — including checking my cal for availability and finding requested attachments / links

4. A specialist for each lane: curriculum, engineering, coaching, hiring, content, ops, and one for every single piece of software

5. Routines that keep running while I’m offline (e.g. monitoring Sentry errors in our apps and proactively fixing things)

6. Group rooms where 2–4 bots share one project thread instead of me copy-pasting context

7. Cloud coding agents that pick up Linear issues and open PRs after running the list of open work by me EoD — then squash-merge to main when it’s done

8. Meeting prep briefs pulled from Granola + Notion before I walk in

9. A growing shareable knowledge base in Notion + a GitHub repo that we update daily based on what happens at school

10. Student progress look-up across Expertise, Followers, and CoFounder without inventing numbers — chat anytime to see where a student is on their business work

11. Mentor Mind that coaches me on how to hold the bar without inventing doctrine

12. Todoist as a central task list where it logs things it’s blocked on for me, or from meetings / emails — and I can paste links into chat to direct it how to solve them

13. Engineering work is automatically tracked in Linear so my and the product teams’ bots don’t collide with each other

14. Presentations spun up in Gamma / Claude Design without me opening a slide tool

15. Plaud / live capture → notes the bots can actually act on

Probably more but these were the immediate ones we thought of.
Nat Eliason @nateliason
GrokBot feels like absolute magic at this point, a meaningful leg up on my previous OpenClaw etc. setups.

And with how easy it is to setup, there's really no excuse now.
♥ 2.1K · ⟲ 207 · 👁 522.0KView on X ↗