The latest blog posts to take off on Hacker News. 6,329 posts.
“Don’t use chatbots as search engines” was great advice for several years... until it wasn’t. I wrote about how good OpenAI’s o3 was at using its Bing-backed search tool back …
Best of both worlds: Building an AI gatekeeper that knows exactly which tests to run for your code changes
All the same privacy harms with online tracking are also present with AI, but worse.
OpenAI’s new research explains why language models hallucinate. The findings show how improved evaluations can enhance AI reliability, honesty, and safety.
Listen in on Jane Street’s Ron Minsky as he has conversations with engineers working on everything from clock synchronization to reliable multicast, build systems to reconfigurable hardware. Get a peek at how Jane Street approaches problems, and how those ideas relate to tech more broadly.
A complete guide to agent architecture, orchestration patterns, trust strategies, and adoption plans for PMs building AI agents.
The most powerful AI platform for enterprises. Customize, fine-tune, and deploy AI assistants, autonomous agents, and multimodal AI with open models.
A big nerd debate with bigger implications for the future of work, technology, and the economy
78% of developers claim AI makes them more productive. 14% say it's a 10x improvement. So where's the flood of new software? Turns out those productivity claims are bullshit.
Alex is the ultimate tool for iOS and Swift app development, empowering developers with AI for Xcode to streamline workflows, tackle complex coding challenges, and boost productivity. Discover what makes it an essential asset for modern app creation.
Our lab investigated whether frontier models can write optimized GPU kernels for Apple devices to speed up inference. We found that they can: our AI-generated Metal kernels were 1.24x faster across KernelBench v0.1 problems, and 1.87x faster across KernelBench v0 problems.
Visualizing the internal state of a Transformer model
From the Zed Blog: You asked, and here it is. Use Claude Code in public beta directly in Zed, built on the new Agent Client Protocol.
We uncovered 1,100+ exposed Ollama LLM servers—20% with open models—revealing critical security gaps and the need for better LLM threat monitoring.
A quick refresher on the maths behind LLMs: vectors, matrices, projections, embeddings, logits and softmax.
This started as an internal Sanity workshop where I demoed how I actually use AI. Spoiler: it's running multiple agents like a small team with daily amnesia.
Anthropic has completed a $13 billion Series F led by ICONIQ, co-led by Fidelity and Lightspeed, valuing the company at $183 billion post-money.
Embedding sizes and architectures have changed remarkably over the past 5 years
Anthropic's August 2025 threat intelligence report on AI cybercrime and other abuses.
Frameworks for LLM Agents have been surprisingly disappointing. I want to offer some principles for building agents based on our own trial & error, and…
AI companies are causing a content drought that will eventually starve them.
Fastly’s survey shows senior developers trust gen AI tools enough to ship 2.5x more AI code, while juniors stick to traditional coding and caution.
Andy Bell had heard companies were forcing employees to use AI tools so spoke to developers to determine if that was the case and learned some horrors in the process.
A quick note on the importance of reading the terms of service!
Neural networks are more useful when placed in a suitable, specialized environment.
From Multi-Head to Latent Attention: The Evolution of Attention Mechanisms What is attention? In any autoregressive model, the prediction of the future tokens is based on some preceding context …
SynthID is a tool to watermark and identify AI-generated content, helping to foster transparency and trust in generative AI.
DeepSeek is a popular open-source large language model (LLM) praised for its strong performance. However, its large size and unique architecture, which uses Multi-head Latent Attention (MLA) and Mixture of Experts (MoE), require an advanced system for efficient serving at scale.
I didn't misclick any buttons. My best wasn't good enough. I'm not good enough.
