Chris Kluis retweeted
The @expo Exact team buried the lead. It’s not about startup speed.
I’m 99% through porting an app I’ve spent 3.5 years building, in 4 days with Expo’s Exact. Faster, better-looking, easier to iterate on. Cost $1000 of tokens ~7M output tokens (compared to the 140M for the C compiler rewrite).
I’m using it against our production APIs now. We’re dogfooding with it. 🤯
The main takeaway is that we built our developer tools around human clock speed. The Exact team are showing what happens when you rethink that from first primitives.
The combination of clock tools, css and contracts as a ui language fundamentally is the unlock. Exact does what it says, it gives you exactly what you asked for and usually better.
Unfortunately, I’m running out of engineering excuses to avoid joining the marketing team now.
Chris Kluis retweeted
Since our pitch deck was uninformative, let me try to explain our whole thesis. Sorry if any part of it sounds obvious.
I argue that any reasonable definition of AGI must imply a system capable of learning as efficiently as a human. LLMs obviously don't pass this bar. When you talk to them, they don't learn anything, since the weights don't change. So, unless true infinite context lands, or GPUs become 10000x cheaper, the current methods won't scale to AGI, at least in that sense.
(Note that this has nothing to do with raw capabilities. Perhaps Fable 10 will be building Dyson spheres, yet still be unable to internalize new knowledge as efficiently as humans do. So, one might ask if we even *need* AGI in the sense above. I think we do, if only because a drastic drop in pre-training costs would make AI more democratic, and prevent one of the worst outcomes in my mind: over-concentration of power.)
As you know, modern AI is based on neural networks. What not everyone knows is that NNs are just very, very slow programs. Nobody in their right mind would sort a list using NNs, because that'd be millions of times slower than just using C. So, why do we implement *intelligence* with NNs?
As a thought experiment, imagine if we could take Fable's weights and distill them, not into a smaller model, but into a raw C program that was observationally identical to it. Literally, a C file with astronomically complex, tangled procedures that, when compiled and executed, behaved just like Fable 5.2.
If such a thing were possible, then we could have Fable at a million tokens per second. And here's the cool part: such a thing *is* possible. At least, as far as computer science is concerned, nothing says it isn't. So, why don't we do just that?
Because we don't know how to *train* a raw program.
Neural networks are slow, but they're differentiable, so we know how to train them. Step by step, slowly, but surely. Raw programs are fast, but we have no methods capable of training them. So we don't. We just stick to NNs, and accept that it takes $50 billion to train a model. We take that as "normal". But is it?
In theory, it shouldn't be like that. As the old saying goes: first you make it work, then you make it fast. We know AI works. Now, we will make it fast. That's the premise behind Symbolic AI. HOC's first milestone is to train a model comparable to GPT-2, with no NNs. If that works, we might have the fastest inference engine in the world. Not by 2x, but by 1000x.
So, in a way, HOC is like DeepSeek or Jev, in the sense that we're all about efficiency, but in a much more extreme way. People are aiming for hardware, but we believe there are still 1000x gains to extract in the software layers, if we're willing to rethink the whole stack from literal scratch.
Many laugh and say this is impossible, but they say so with no basis. They just observed that many failed to build symbolic AI models in the past, and concluded it is impossible, or very hard. That's not a formal proof. In fact, there are many reasons people in the past could've failed, yet the actual problem may end up being easier than most anticipate. Symbolic AI flourished half a *century* ago. Back then, they didn't have:
- As much knowledge as we do
- Massive GPU clusters
- Methods to scale Symbolic AI (Bend3, HVM5!)
So, just like everything that once seemed impossible, there's one core prerequisite to making it possible: *trying*.
That's what we'll do!
I don't think I'm smarter than all the brilliant researchers who worked on this problem in the past, and failed. I just think I'm equipped with tools, compute, and knowledge that they didn't have. So, standing on the shoulders of giants, and with enough dedication, I don't think it is so unlikely we stumble upon some symbolic algorithm from which, given enough compute, unexpected capabilities emerge.
If it does work, I do think business is straightforward. At the very least, we would be able to sell incredibly high quality datasets to big labs, but, at that point, just selling inference might work as well. Business is obviously not the wall, it is all on whether such massive software-based speedups exist, and whether we're able to discover them.
Company harnesses that are both the:
Context management engines AND
“where you work”
Will win.
I think Buzz, Orbit, Macro are beginning to show that, but it’s kind of shocking OpenAI isn’t doing more of this in Codex directly.
Next step is dynamic ontology-based on graphs
26
Chris Kluis retweeted
Here's Campfire in Goose.
It serves pages
- 1.2x faster than the Jai implementation
- 6.9x faster than the Rust implementation
Live messages to 64 readers: 1.4x faster than Jai.
At 10,000 readers: 38 MiB vs Jai's 651 MiB, p99 0.25 s vs 15 s.
github.com/kaitk/once-campfi…
A new tiling system for macOS
1. Apps in columns
2. Clickable icon for scrolling
3. Multi-monitor support
4. Built in performance monitor
5. MCP using Apple Intelligence for layout (native)
ckluis.github.io/kolom/
Chris Kluis retweeted
turns out bend was ahead of the curve
New on our Frontier Red Team blog: using Claude to formally verify software at large scale.
Formal verification can prove that code is free of certain bugs, but historically it has only been feasible for small codebases (or required a lot of expert time).
With AI models, however, this is beginning to change. In collaboration with @theoremlabs, we used Claude to prove security properties in over one hundred real-world open-source libraries over the span of a few weeks.
Read our research in full here: anthropic.com/research/forma…
Chris Kluis retweeted
If you just read on this this week, let it be this blog by Ant and Theorem Labs. Traditional software testing can not capture system behaviour for each an every scenario that formal verification can.
Quote: "Claude also discovered bugs that had eluded detection for years, despite trillions of test runs. As model capabilities continue to improve, we expect formal verification to become an essential part of secure software development."
New on our Frontier Red Team blog: using Claude to formally verify software at large scale.
Formal verification can prove that code is free of certain bugs, but historically it has only been feasible for small codebases (or required a lot of expert time).
With AI models, however, this is beginning to change. In collaboration with @theoremlabs, we used Claude to prove security properties in over one hundred real-world open-source libraries over the span of a few weeks.
Read our research in full here: anthropic.com/research/forma…
Chris Kluis retweeted
Let me put my TypeScript hat back on for a minute
Content mappers look outrageously exciting. This is now possible:
1. import any non-TypeScript file (markdown, vue, svelte)
2. Write a content mapper to extract types from it
3. Run tsc with --runExternalCode
4. You get type inference from NON TYPESCRIPT CODE
The simplest case:
1. Import a markdown file
2. Add a bit of config to tsconfig.json
3. Run tsc with --runExternalCode
4. Get fully typed markdown frontmatter
Unbelievably useful, a dreamland for tinkerers
TypeScript 7.1 Beta is now available! Now with a preview API, content mapper support, new features, and some great optimizations. Give it a try and give us some feedback! devblogs.microsoft.com/types…
Product KPIs, scored live 📊
Drag a guardrail. Every KPI Status re-scores.
• 4W / 13W / 26W views
• Promo week drill-down
• Per-market KPI map
• Trend vs prior period
• Nested filter groups
• Break-even as you type
• Full details sheet
Built on @shadcn + @tan_stack Table
Chris Kluis retweeted
iPhone-style folders for your Mac Dock. Group apps, shorten the Dock, keep every app one click away.
Chris Kluis retweeted
Introducing Pixel - a library for building GUIs in the terminal
- 120 FPS pixel-perfect graphics
- Embeddable inside existing TUIs
- Built in chromium webview component
pixelgui.com
Experimenting with window/tiling inspired by Paneru & niri.
Thoughts?
ribbon-desktop-concept.cklui…
39
Chris Kluis retweeted
nah bro we're in a simulation for sure
This entire ocean is a 61 KB HTML file.
Built with Claude Code + Opus 5.5 in about 30 minutes, according to its creator.
It runs in the browser using WebGPU: FFT waves, whitecaps, sun glitter, a physically based sunset sky with clouds, and a free-flying camera. No game engine or rendering library.
Claude rendered screenshots in headless Chrome, then had separate subagents critique them. The creator gave feedback along the way, mainly on the clouds.
The 61 KB build is minified and self-contained. Download it, open it in Chrome with WebGPU support, and explore with WASD.
💡for @Apple
Let me install an app 2x on a duo to run it 2x next to each other.
Claude Personal.
Claude for Work.
60
Chris Kluis retweeted
Source code is not available yet; it will be available in the coming weeks. Launching the website now to get the project qualified for @AnthropicAI [1] Glasswing
overalls.dev/
[1] 🐜 folks, hit me up in DMs.
Chris Kluis retweeted
Introducing TrellisWM:
The Fractal Window Manager for MacOS
Its how the desktop should have always been.
Download at
TrellisWM.app






















