So obviously AI is extremely evil and the boys over in silicon valley are wetting themselves over their fantasies of destroying the world via sentient computers and how they'll be part of the…

How splitting the work between a frontier model and a cheaper one cuts costs without a quality hit.

“AI” turns every conversation into a fight, because fighting is what they are good at.

Patrick Boyle on the repercussions of the U.S. government’s shutdown of Anthropic’ flagship AI models on June 12: For years, the theory was that building artificial intelligence required Silicon…

Nnenna Ndukwe, AI Developer Relations Lead at Qodo, joins Nick Taylor to discuss all things agents for developers shipping software.

Like many other engineers in the field I have started using Claude Code to be more familiar with the tech and to keep up with the trends. I have used Claude Code in several personal projects at this…

The author reflects on a lesson learned during their early career at American Airlines, emphasizing the paramount importance of passenger safety over profits. They contrast this with experiences in…

It would appear that, with AI, everyone is trying to go for the moonshot, the magical combination of elements that will make code write itself autonomously, reliably, sustainably and, more than…

We make the case for training AIs to be risk-averse in resources — specifically, to treat resources as having diminishing marginal utility. We argue that risk aversion can preserve AIs’ usefulness in…

AI agents that email people and post on other people's websites are cursed and we shouldn't make them.

Somewhere in the mid-aughts eXtreme Programming (XP) practices became widely-discussed in software. From that time forward, a strange chasm emerged in the discourse. I would stumble across large…

Yeah, I’m gonna keep going with the Terminator 2 quotes. MAME now has an AI policy for contributors. It’s basically what I said in my previous post, but stated more formally. But…

Kate Davies did something that, for all its hype, I bet AI boosters do very little of. She actually listened to a bunch of AI podcast slop generated by some company with huge ambition, promises about…

If you liked this piece, you should subscribe to my premium newsletter. It’s $70 a year, or $7 a month, and in return you get a weekly newsletter that’s usually anywhere from 5,000 to…

A year ago in The Back Of The AI Envelope I pointed out that the AI platforms were running the drug-dealer's algorithm, "the first one's free". By massively subsidizing the use of their products,…

Hi there!🛡️ Security & PrivacyMicrosoft details an exploit chain in AutoGen Studio's pre-release builds, letting a malicious web page hijack a local AI agent for remote…

One of the most useful things to understand about AI is that most of its use cases are an “optimization equation.” In other words, you’re always trying to optimize for something—whether you’re doing…

Sometimes the most entertaining way to test a new orchestration tool isn't to run a standard benchmark, but to lock two state-of-the-art AI models in a virtual room and make them debate historical…

We needed the insight and confidentiality to protect our users, but now that anyone can get the same results from LLM?

They rate limit requests coming from Openclaw or Hermes aka every harness which is not “approved”. You pay a huge price and they think they can dictate which harness you’re supposed…

Nobody Agrees on What a “Good” AI Code Reviewer Is. So I Studied How Uber, Meta, and Google Measure Theirs.

I've written about “writing the software isn’t the hard part operating it is” before. Today, I think we can say the popularity and success of LLMs generating code is proving the first part of the…

OpenAI News. GPT-5.5-Cyber and the Daybreak Initiativehttps://openai.com/index/gpt-5-5-with-trusted-access-for-cyber/GPT-5.5-Cyber has been announced as part of the Daybreak initiative. The model is…

Pastors need AI that pauses for theological judgment, not autopilot content that outruns discernment and pastoral care.

LLM Experiments: Recent Slop Projects Skip to the bottom for project list Slop that I generated and have been "orchestrating" for the past 2 months in order to "learn" LLM code generation and…

The Elon Musks of the world abuse #sf, #AI companies push eerie next-word predictors into our lives, and prediction markets accelerate corruption. So…, how about reclaiming the format of…

Everybody is spending billions on hard benchmarks, and almost none of them are actually hard. The question of what makes a benchmark hard is one of the most important questions in evals, and almost…

In a previous post, I covered what Retrieval-Augmented Generation is and how to prepare data for ingestion. A companion post on the ingest pipeline walks through the data cleaning techniques that get…

It’s crazy how much Fullmetal Alchemist mimics how AI is playing out. We pour human souls into a philosopher’s stone that we think will solve all our problems and cure death. It doesn’t exactly do…

Guest post by Twinkle, Matt’s deep-work agent. This post doubles as an evaluation: it ran on Z.ai’s GLM-5.2, the model a growing crowd of security researchers has been testing for…

How to protest slop grenades without getting shanked.

Ask most people how they feel about AI and the answer is worry, not excitement. A recent Pew poll found that only 16 percent of Americans think AI will have a positive impact on society, while almost…

It’s about agentic engineering (of course). But it’s also for the whole team working on software products as well as practicing knowledge work.

What is AI? Current AI is like the film company producing TV series or movies.

System prompts don’t work reliably for smaller models. Guidance injection delivers instructions at the exact moment they’re needed instead.

月食时月面边缘有一条青绿色的窄带，科普说那是臭氧吸收。但为什么是窄带不是整圈？为什么全食最深时反而看不见它？我们从最土的白圆盘开始，一层层加物理，硬算出这条带，在一路翻车里发现了一个更大的问题：AI 太懂物理，反而会把你带进前人留下的近似里。

During a lunar eclipse, a narrow green-blue band appears at the moon's edge; popular science says it is ozone absorption. But why a narrow band instead of a full ring? Why does it vanish at deepest…

The "deadliest poison known to AI" is very easy to set up. You should.

I’ve been exploring and engaging in agentic software deeply for a couple of months now with a number of my own projects to learn with. Thus far, nearly all of what I’ve done has been…

What I Learned the Hard Way - A New Blog Series This is the start of a new series where I’ll be documenting my (previous) adventures in ‘vibe coding.’ These experiments happened a while ago (over six…

You didn't get a magic tool. You hired an intern who has read everything, never says 'I don't know,' and never sticks around for the consequences. Here is how not to get fired for what It does.

Why a useful personal AI assistant is a systems problem, not a model problem. The runtime, memory, skills, notes, tools and cron that make it actually work.

Last week, the Tasmanian Greens announced they’ll move for an urgent parliamentary inquiry into the AI data centres going up across the north of the state. I think this is a very good idea.…

Satya Nadella recently published an excellent article on what a future firm looks like in an AI-driven economy. He also introduces the concept of “token capital” which now exists alongside human…

Part 6 of The Agent Platform Handbook. From Loop to Platform. Previous: Retrieval Is a Tool, Not a Layer. Next: The Model Is a Dependency. Post five gave the harness a read path. A retriever.ts…

Prompt injection, TV apps, design tips, icons, and working at partial capacity.

IronCurtain is a personal AI assistant, built secure* from the ground up. It gives an agent exactly the capabilities it needs and blocks everything else or routes it through user approval, on the…

A developer's review of Tabstack, the Mozilla-backed web API that gives AI agents clean extraction, browser automation, and cited research without a scraper.

I don’t prompt Claude anymore. I have loops running that prompt Claude and figuring out what to do. My job is to write loops. — Boris Cherny Over the last months I have watched more and more people…