@alexgraveley Yes. If you haven’t bought a shaver recently (many men just use the first one they ever bought) this one will run you $40 but save you so much of your time
Perspective matters here. He’s building Claude which has billions in revenue rolling in. He’s incentivized to do it using Claude because that’s the product and they need to dogfood it. A few thousand dollars a day in tokens is the cost of doing business. You could argue the output is shit but the revenue says the market wants it
The real fun is the models tend to gaslight you into thinking you’re building this incredible piece of revolutionary software that needs to be canary deployed and have thousands of tests for every small piece of it.
You’re building a website bro.
It’s actually simpler than that. We’re building fundamentally better systems and infrastructure.
Why? We understand it better and now tools can explain it better to others. You can take this new approach in two ways—build lazier and worse or use it to build robust and well thought out complex systems that are better long term.
The catch? Most companies don’t need that level of rigor and they reward the former lazy behavior because those guys are shipping at incredible speeds (look at any AI lab)
Finally—the human factor. If you have rigor and the tools to build better you have to convince others it actually is better. They have the same tools to form a rebuttal. So we are stuck in constant debates and “analysis paralysis” sets in.
It’s actually simpler than that. We’re building fundamentally better systems and infrastructure.
Why? We understand it better and now tools can explain it better to others. You can take this new approach in two ways—build lazier and worse or use it to build robust and well thought out complex systems that are better long term.
The catch? Most companies don’t need that level of rigor and they reward the former lazy behavior because those guys are shipping at incredible speeds (look at any AI lab)
Finally—the human factor. If you have rigor and the tools to build better you have to convince others it actually is better. They have the same tools to form a rebuttal. So we are stuck in constant debates and “analysis paralysis” sets in.
What is a “quant”?
It’s been a question rolling in my head for a few years—a quantitative. A profession reserved for the elite engineers from the top schools. They command high salaries from the likes of high frequency traders and hedge funds—Citadel, etc.
But here we are in 2026 and the barrier to entry has been lowered by—artificial intelligence. So follow me on this journey if you’d like to learn more.
@FinancialPhys Not to mention most of the job postings online are fake and posted externally for a period of time as a legal requirement even if there is an internal hire that they are giving it to
Stumbled upon a Codex skill that creates cool illustrations to explain topics or tell stories.
You feed it text (blog, article, narrative, even code) and it makes explainer graphics with this cute blob character.
I gave it the repo for the X recommendation algo and got this 👇
I think I finally found the best way to interact with agents. Stole some parts from @badlogicgames pibot, mashed it together with @kyutai_labs stt and can finally be the man I've always wanted to be! Fully local on an m4 macbook btw!
If you’re a developer in 2026 try to build your own coding agent. It’s not difficult and those core skills and tools will help you navigate this new world of Claude code, Codex, etc and the tradeoffs these labs are making to build them.
@trikcode Just keep it really simple. Avoid all of the latest and greatest ways to “prompt” or whatever. If it’s really worth it the labs will add it as a core feature.
@zeeg Is this bait?
Serverless has a purpose and is often the best choice especially for services that don’t need high availability. You get compute when you need it and don’t pay when it’s not being used.
What’s not to like?