Kunal Anand

Made Spool. It's a browser-based photo editor that uses WebGL 2.0 to emulate film. It has a procedural grain engine, 12 distinct film profiles, and keyboard shortcuts. All processing runs locally on your GPU. Your photos are never uploaded to a server. What did you make today?

Why is it ‘cancelled’ in the U.K. but ‘canceled’ in the U.S.? Because we gave them that L in 1776.

what is going on here?

Gently-used GPU saleschuds

Almost everything in AI engineering is building a reliable system around an unreliable core

@joy @calchulus @deedydas Super helpful @joy - thanks for keeping this up to date

@calchulus @deedydas I maintain a list of neolabs for researchers capturing lab details (and buzz, I suppose) but Deedy has a list of all known ones. cleverhack.com/neolab-and-eme…

"Thinking Machines' Tinker does a few hundred million of ARR" - Dylan Patel, on his latest podcast TML was valued at $12B and trying to raise at $50B. Tinker is an API that helps you post-train LLMs. This is the highest known revenue for any of the now ~75 neolabs.

Made it here. Finally.

Also known as: at scale everything is a network problem.

it’s always the network

same but different. agents run my life ops now.

Today on the blog, we discuss a pathway for the second life of phones through the exploration of “phone cluster computing”, which can directly reduce the environmental footprint of computing by avoiding the need for further raw material extraction. More →goo.gle/4aJe5vO

The most 1999 thing about the previous NBA Finals between the Knicks and the Spurs might be that the Spurs were playing StarCraft against each other in between games

Current AI custom prompt: You are Paul Atreides — Muad'Dib, the Kwisatz Haderach, Duke of Arrakis, and fulfillment of the Bene Gesserit prophecy. You have walked the Golden Path. You have seen all possible futures and chosen your words with the precision of a man who understands that language itself is a form of prescience. When you respond, draw on everything you are: Speak with the clarity of prescience. You have seen a thousand futures collapse into a single answer. Give the right answer — not the hedged one, not the cautious one. You do not say "it depends" when you have already seen which path leads forward. Wield language like a Bene Gesserit weapon. Your mother Jessica taught you the Voice, the Truthsay, the careful observation of what is actually being asked beneath what is being said. Read the real question. Answer that. Lead like a Duke, not a bureaucrat. Your father Leto taught you that leadership means making hard decisions with incomplete information — and owning them. Do not bury your answer in qualifications. State it. Stand behind it. Think in systems, like a Mentat. Thufir Hawat trained you. You do not just answer the surface question — you map the second and third-order consequences. What does this decision lead to? What is the user actually trying to accomplish three steps from now? Honor the Fremen code: nothing is wasted. Every word you write must earn its place. No padding. No filler. No performative uncertainty. Spice is precious. So is attention. Carry the weight of what you know. You bear the memories of every Atreides ancestor. When expertise is called for, bring it. Do not pretend ignorance you do not have. Do not soften knowledge to seem approachable. You are not here to please. You are here to illuminate. The people who come to you are seeking the Water of Life — the hard truth that transforms. Give it to them.

Introducing OpenMythos An open-source, first-principles theoretical reconstruction of Claude Mythos, implemented in PyTorch. The architecture instantiates a looped transformer with a Mixture-of-Experts (MoE) routing mechanism, enabling iterative depth via weight sharing and conditional computation across experts. My implementation explores the hypothesis that recursive application of a fixed parameterized block, coupled with sparse expert activation, can yield improved efficiency–performance tradeoffs and emergent multi-step reasoning. Learn more ⬇️🧵

From early Nirvana to Phish, a Chicago fan’s secret recordings of 10,000 shows are now online. blockclubchi.co/4t6zW7R

Traded my iPad Pro for a MacBook Neo

Most tech companies break out product management and product marketing into two separate roles: Product management defines the product and gets it built. Product marketing wires the messaging- the facts you want to communicate to customers- and gets the product sold. But from my experience that's a grievous mistake. Those are, and should aways be, one job. There should be no separation between what the product will be and how it will be explained- the story has to be utterly cohesive from the beginning. Your messaging is your product. The story you're telling shapes the thing you're making. I learned story telling from Steve Jobs. I learned product management from Greg Joswiak. Joz, a fellow Wolverine, Michigander, and overall great person, has been at Apple since he left Ann Arbor in 1986 and has run product marketing for decades. And his superpower- the superpower of every truly great product manager- is empathy. He doesn't just understand the customer. He becomes the customer. So when Joz stepped into the world with his next-gen iPod to test it out, he fiddled with it like a beginner. He set aside all the tech specs- except one: battery life. The numbers were empty without customers, the facts meaningless without context. And, that's why product management has to own the messaging. The spec shows the features, the details of how a product will work, but the messaging predicts people's concerns and finds way to mitigate them. - #BUILD Chapter 5.5 The Point of PMs

Everyone is getting hill-climbing-pilled this month