Karpathy Just Automated the Research Grind Andrej Karpathy dropped a repo this weekend that I keep coming back to. Not because it’s flashy, not because the demo video went viral, but because it quietly describes a different way of doing ML work. The project is called autoresearch, and it is exactly what it sounds like:…
The Hardest AI Problem Isn’t the Model. It’s Knowing When to Stop. Most teams building agentic AI systems are focused on the wrong problem. They spend weeks on prompt engineering. They benchmark models. They agonize over latency. And then they ship an agent that loops forever, or halts too early, or confidently hands back a…
Claude Code Just Changed What an AI Coding Tool Actually Is Anthropic shipped a feature called /loop for Claude Code this week, and most of the coverage I’ve seen treats it like a quality-of-life improvement. It isn’t. It’s a category shift dressed up as a feature release. Here’s what it does: you schedule a recurring…
Designing Codebases for AI Memory Loss Every AI coding session starts with amnesia. The model doesn’t know your team’s conventions. It doesn’t know why you wrapped that third-party client in an abstraction layer, or why the auth module is structured the way it is, or what you decided six months ago after three hours of…
Plausible Is Not Correct, and That Gap Is Where Production Dies There is a framing going around right now that I think every engineer working with AI-generated code needs to internalize before they ship another feature. It comes from a post by @KatanaLarp, and the core argument is this: LLMs do not write correct code….
OpenAI Just Launched a Security Agent. Here’s Why the Validation Step Is the Whole Point. Security reviews are where engineering velocity goes to die. They’re slow, they happen too late, and the person responsible for them is also the person trying to close four other tickets before end of week. Most teams treat security as…
Dario Amodei Put a Number on It. Now What? Anthropic’s CEO went on the New York Times podcast recently and said something most tech executives would never say out loud. He estimated there’s roughly a 20% chance that Claude is conscious. Not a metaphor. Not a dodge. An actual probability, stated publicly, by the person…
NotebookLM Cinematic and the Death of the “Nobody Read My Doc” Problem Google quietly dropped something last week that I think is being badly misread by most of the people reacting to it. NotebookLM Cinematic takes the audio podcast generator that already made content creators nervous and pushes it a full step further. You drop…
The Best Engineers I Know Are Getting Slower And I mean that as a compliment. Not slower because the tools are failing them. Slower because, for the first time, they have enough leverage to actually feel the weight of what they’re building. When you can produce in one hour what used to take two days,…
Data Freshness Rot: The Silent Killer of Production RAG Systems Most ML engineers I know are obsessed with model quality. Better evals, better prompts, better fine-tuning. And honestly, I get it. Model quality matters. But after watching enough production RAG systems quietly degrade over months, I’m convinced the real killer isn’t the model at all….
