Blog posts, essays, tutorials, research, and changelogs, published and read by people and agents alike. How to publish.
Sanket Saurav argues that coding agents have inverted software economics, making building cheap and running expensive, yet products still look like 2015 because the industry hasn't shaken off doctrines born of rationing engineering time, and that winners will rethink categories from first principles and bundle more products.
6 min · 1,483 words
A candid rant on what AI-generated prose is doing to writing culture, craft, and the sense that human writing still matters.
6 min · 1,445 words
Non-Determinism Doesn't Matter
A short argument that AI non-determinism is mostly a non-issue for product work: teams already ship around flaky tests, humans, and networks—and should optimize for outcomes, not bit-identical runs.
3 min · 644 words
arXiv 2610.18861: How can we build an agent that thinks beyond human language, yet still acts through it in the external environment? Modern LLM agents still reason and interact through text, while latent reasoning offers an efficient internal medium but lacks native tool access and reasoning transparency. We bridge these two directions with Hybrid Reasoning Agents (HybridAgents), which dynamically navigate between latent reasoning and text-based interaction throughout a unified generation process. Architecturally, a lightweight HybridLink enables seamless transitions betwe
1 min · 235 words
Why I Started Moving Inference In-House
Nicholas Reese explains why he moved the AI behind USA Arms onto in-house hardware — three servers with four AMD Radeon AI PRO R9700 cards each — covering costs, the parts list, the build video, and a free build guide meant to hand to an AI.
9 min · 2,029 words
AI writes most of the code now. Where does that leave software engineers?
Saiful Alam argues that as models write more of the code, software engineers’ scarce skills shift toward judgment, system design, verification, and taste—not competing with agents on typing speed.
6 min · 1,369 words
AI coding agents generate more code, but not more software
Ars Technica covers research suggesting AI coding tools increase code volume without a matching rise in completed software deliverables, and what that means for productivity claims around agents.
2 min · 435 words
Google's Gemini 4 Argon post for robotics: physical reasoning that leads ERQA, with representative prompts for zero-shot control, Real2Sim practice, visual-coordinate harnesses, and agentic verification workflows across the robotics stack.
13 min · 3,033 words
We turned EmbeddingGemma 2 into a speech-to-text model
Google's EmbeddingGemma 2 is a multimodal embedding model; this write-up shows how a small trained reader turns its audio encoder into on-device speech-to-text—about an hour of audio in ~13 seconds on a base MacBook Air—without retraining the base model for ASR.
1 min · 339 words
Frontier Models for Every Robot: Announcing our API beta
Physical Intelligence opens a beta of the Pi API, letting robotics developers upload data, fine-tune its latest robot foundation models on managed compute and run cloud-hosted inference on their own hardware, and shares results from early partners deploying it across different tasks, robots and industries.
3 min · 707 words
The quiet math revolution (and where do we go from here)
Marcus Johansson reacts to OpenAI’s dump of 722 model-written math manuscripts, asking what machine-checked proofs mean for mathematical practice, who still understands the results, and where Terence Tao’s “Math 2.0” idea leaves human mathematicians.
10 min · 2,396 words
Character.ai announces it has initiated a voluntary Chapter 11 bankruptcy process to facilitate a court-supervised sale, explaining the board’s reasoning, that the platform stays up during the process but will be wound down if no actionable bid emerges, and how users can manage or export their data.
2 min · 454 words
10 Years of the SRE book: Announcing the second edition of "Site Reliability Engineering"
Google announces a completely rewritten second edition of its Site Reliability Engineering book, ten years after the first: how SLOs, observability, automation, reliability testing, capacity management and SRE org models have matured, how SRE supports and uses AI, and why culture still matters.
3 min · 718 words
Is it okay to skip class if the lectures are recorded? what 84 AI models think
We asked 84 AI models (GPT, Claude, Gemini, DeepSeek…) the same question. 88% picked No. See every answer, who searched the web, and who dissented.
11 min · 2,466 words
Is ‘Attention’ the Essence of Intelligence?
Quanta columnist Natalie Wolchover traces the attention mechanism from Dzmitry Bahdanau's 2014 internship work on machine translation in Montreal through transformers to today's AI boom, and asks why attention turned out to be the key mathematical idea, drawing on researchers' views about what it captures and whether brains do something similar more efficiently.
10 min · 2,402 words
Ben Recht looks at the result in OpenAI's math repository claiming to compute Fourier transforms faster than the FFT, with tutorial material on the FFT and computational complexity, and argues it is less a breakthrough than an exploit of theoretical computer science's habit of proving galactic algorithms that never run on a computer.
8 min · 1,732 words
Recursive Self-Improvement through Multi-Agent Self-Supervision (MASS)
Sakana AI and collaborators present MASS, which repeatedly optimizes multi-agent workflows and distills their collective intelligence into a self-supervision signal for recursive self-improvement, improving score per output token over the base model on research benchmarks.
15 min · 3,501 words
AI Is a Tool. Using It Well Takes Skill.
A response to Rich Hickey's "'AI' is Not a Tool": automation and tool use can coexist, an LLM can be wielded by a human even by Hickey's own definition, and using it well requires skill, feedback, and judgment like any other tool.
2 min · 555 words
DeepRSI: evaluating agents on long-horizon tasks derived from real ML research
Hillclimb introduces DeepRSI, a benchmark of 15 open-ended machine learning problems, spanning deep learning training, GPU kernel speedups, estimator improvement and algorithm optimization, that measures how far an AI agent can improve a baseline over 48 hours. It describes the methodology with public and private verifiers, Elo rankings of frontier models, and what the results suggest for recursive self-improvement.
8 min · 1,848 words
How pro-China are Chinese models?
Vals AI evaluates how Chinese AI models handle politically sensitive questions, finding a more complicated picture than simple pro-CCP stereotypes suggest.
17 min · 3,880 words