Aryan

It is entirely possible for a statement to be false and still function as good advice. Good advice is good simply if believing it reliably pushes you towards a better state of being (whatever that means for you). That's it. No precise accordance with reality required. Consider @pmarca 's takes on introspection. The correctness of his stance is debatable, to say the least. But for profiles that are highly susceptible to getting nerd-sniped / stuck in analysis-paralysis, it's pragmatically excellent advice.

@jakubkrehel great post

Understanding Gradients jakub.kr/work/gradients

As models improve, the value of harnesses will asymptotically approach ~0.

@tenobrus you should create a council with mythos 5 + gpt 6 + gemini 4 and take the majority vote for better accuracy

god i can't wait

pretext is an impressive feat of engineering, but I haven't seen a single demo that materially improves UX/UI in a practical use case (yet). It appears to mostly be a perf gain, which is of course welcome, but not some earth-shattering reinvention of the frontend as we know it.

The ultimate irony of the current state of AI<>design is the utter lack of creativity. Tens of millions are being poured into cookie-cutter agent canvas clones that invoke the same LLMs and provide little to no innovation on the interface or core primitives of the design process.

When you force AI to use heuristics like "no gradients" or "no em-dashes", you are filtering out a sizable chunk of the possibility space. Most of this chunk is (presumably) bad/undesirable, but there are nearly always diamonds in the rough that end up being collateral damage. The result is an increase in mean output quality at the cost of slightly less stylistic freedom. As you stack filters, this lack of freedom compounds sharply. The result, for creative fields at least, is outputs that feel template-like and sterile. Never wrong or ugly, but never great either. Just alright and passable. A distribution that has tightly clustered around the 70th percentile. This is why I firmly believe that design excellence is not a harness problem; it's a model problem. Encoding the complex, jagged, and discontinuous boundaries between good and bad design is a job for parameters, not system prompts or tool calls.

@fkasummer agreed. venture capital will become less about intuition/asymmetric bets and more about applied math/brute-forcing the entire hypothesis space. analogous to what jim simons did to wall street. every market will become an efficient market.

Eventually, most computation will be autonomous. So will most of the economy. We will have autonomous companies. (Nothing to do with LLMs.)

@bykahlil Tailwind's step-function utility classes are like training wheels: helpful when you're starting out, but meant to be outgrown!

I wrote our marketing site in vanilla CSS. No Tailwind. No random frameworks. Curious yet?

I once thought so too, but I increasingly believe that this is a misconception, usually caused by mapping the human brain's limitations as a 'rendering engine' to LLMs -- unless, of course, you refer to something other than vision capabilities in your usage of the word "Sight". Code is lossless and a precise source of truth, while PNGs are lossy and resolution-capped. I find it highly unlikely that the latter allows for better generalization.

To acquire Taste, Agents first need Sight.

Whether or not you'll endure is determined less by how much you're suffering, and more by what you think you're suffering for.

There is a certain parasitic force that almost inevitably spawns and feeds on any institution that achieves considerable success, like a poisonous bacteria that refuses to die before its host. The force I speak of is, of course, prestige.

just spawned a new conway automation, pasted a paul graham essay into its genesis prompt, named it "elon musk" (nominative determinism), and gave it 200 usdc

@patelpurav05 Lfg!

we just launched clayzo clayzo gives product and design teams a faster way to prototype ideas, gather feedback, and hand off changes to engineering in the context of the real product. instead of relying on docs, slack threads, looms, tickets, and waiting for deploy links, teams can move through product iteration in a much more direct way. we’d love to hear your thoughts!

@bykahlil @naturalpay lfg

210 days ago, @naturalpay was just a one-pager and a memo. Today, we’re coming out of stealth. Natural is the agentic payments platform powering frictionless money movement between agents, businesses, and consumers. Wallets. Payments. Ledgering. Routing. Identity. Compliance. Credit. Observability. Risk. Everything needed to move money. Engineered for agents and designed for humans. These primitives give you the ability to transact without becoming a payments expert or stitching together a dozen fragmented tools. Huge thanks to our team of 10 (soon to be 25), our early investors, and the supporters who believed in this vision from the start. If you want to help shape how money moves over the coming decades, we’re hiring. And if you’re building agents you should probably be moving money too. Reach out. Read more about Natural and the products we’re launching in the blog below. natural.co/blog/introduci…

I am really excited about x402 and agentic commerce in general. But there's this braindead, hype-induced misconception floating around that the economy is going to somehow explode with 10x more throughput/activity because now agents can buy things. As if there will be "more money to be made." That's just completely wrong. AI agents will only spend insofar as that spending serves the human/institution that owns them. Value exists only if it can eventually be transferred back to the world of atoms. Agentic commerce doesn't inherently expand the pie, it changes the interface. Which is meaningful in its own right, but not an economic explosion.

@Austen definitely more than that x402scan.com

It’s weird because yes the future seems to be selling to agents yet how many AI agents on the planet currently have purchasing power. <100?

@TheStalwart This was always possible, no LLMs required. Mathematics is the only fallacy-proof language.

The fact that one can get an LLM to put together a cogent-sounding argument in favor of virtually any position has left me feeling cynical about human debate and discourse. That it’s (almost) all a game wholly sealed inside language itself without tether to the world.

A great alternative to the mom test is just asking your younger brother.

To grow up is to overcome your shiny object syndrome.