In-depth guides, frameworks, and analysis covering the AI landscape — from security fundamentals to emerging applications.
A one-layer transformer you can take apart in the browser. Attention, the MLP, a logit lens, sampling and ablations, all computed live from weights trained for the purpose, and what those weights turned out to contain.
Read more →The first piece said nobody had yet specified an AI flight recorder. It was wrong: three proposals were already public. The recorder is turning out to be the easy part. What is still missing is who holds the record, who may open it, and what it may be used for.
Read more →The Oireachtas heard last week that data centres now use nearly a quarter of Ireland's electricity. The debate runs in megawatts per site, and no small business can do anything with a megawatt. The number they need is how much energy one answer costs. I've been measuring that on my own machines, and the biggest lever wasn't the model.
Read more →Xiaomi has released the highest-scoring open-weight model anyone has measured, under the MIT licence, and almost nobody can run it. The weights are the least useful part of the release. The recipe, the graded environments and a 9B student model are the parts that fit on a desk.
Read more →A new study finds AI tutors match expert human tutors on GRE learning gains, one of them at a nine-hundredth of the cost. The result is real. The mechanism it reports — faster replies, more messages, more practice — is also the shape of an engagement loop, and we already know what fluent machines do to people's willingness to check.
Read more →The Titanic carried more lifeboats than the law required. The law had stopped counting at ten thousand tons. What followed was not a ban but a way of learning from accidents: a recorder, a duty to report, and an independent investigator. AI has two of the three on paper, none in force, and one of them built backwards.
Read more →Generative AI reached 62% of American adults in three and a half years. The internet needed eight. The climb rate is about the same. What differs is where each one started, and who's paying for it.
Read more →Todd asked whether it's a mistake to sort a system like Claude.ai into intelligent or not-intelligent. It is — but not for the reason people usually give. A tour of the add-ons that make the thing you're talking to, and why the yes/no question can't even find its referent.
Read more →People keep asking me which AI is best. The question assumes an oracle — one machine you can settle on and stop thinking about. That's the exact habit that gets you burned.
Read more →Anthropic just closed $65B at a $965B post-money valuation. That's not a refutation of the bubble thesis. It's a date stamp on the next test of it.
Read more →Kahneman split thinking in two. The standard worry about AI is that we'll stop using System 1. That's not what's happening. Not to me.
Read more →A backgrounder on the agentic pipeline behind the FoxxeLabs stack — the loop that turned one new project every four months into 2.67 a week, explained for developers and founders alike.
Read more →We built a full cognitive architecture modelled on the human brain — from biometric homeostasis to embodied swarm intelligence — using Irish-language names, open-source tools, and a theory that memory is topology, not storage.
Read more →Fifty years of software development, an AI coding partner, and the question of what the human is actually for.
Read more →Since the first Léargas experiment, the corpus grew from 33,440 to 56,006 documents, new data sources joined the manifold, and we built three interconnected visualisations — including one that shows a writing career and a cognitive history on the same animated canvas.
Read more →An independent path to the same architecture — persistent memory, direct repository access, and a strict division of labour. I got there second, and the parallel is still uncomfortably exact.
Read more →I built a tool that runs two independent Claude sessions side by side, lets them argue, and lets you fire one's answer at the other. It's the most useful AI tool I've made.
Read more →How persistent memory, direct git integration, and a strict architect/coder split produce software that neither human nor AI could build as well alone.
Read more →From empty repo to GPS-tracked life trail in one session. How we built a personal worldline data collection system with FastAPI, React Native, and a glowing map of Ireland.
Read more →It started with a command-line hack to change the music. Eighteen months later, Claude Code is generating half a billion dollars a year and redefining what software development looks like.
Read more →On growing minds rather than programming them, on software that fits like a handmade shoe, and on what a glowworm moving through space-time has to do with the question of what intelligence is for.
Read more →We gave the temporal manifold a body — blobs that grow and fade with sound, Ebbinghaus decay, and a decade of an author's publishing life flowing in from 474 Google Sheets. Here's what emerged.
Read more →What if your memory system wasn't a database? We replaced records with a probability distribution over semantic space — a Gaussian Mixture Model that deforms when you think, consolidates while you sleep, and makes connections you didn't program. Here's what we built and what we found.
Read more →A self-contained log page for the Léargas memory map — topic blobs rising and falling across three years of conversation history, now with a full Web Audio sound engine. Sound on.
Read more →AI isn't just automating work — it's compressing the distance between intention and execution, activating latent ambitions, and shifting the real barrier upstream to judgment and taste.
Read more →We built a consolidation layer that runs on top of Mnemos and Radharc to synthesise cross-domain memory bridges — the equivalent of what sleep does to human episodic memory. Here's what we found.
Read more →Most developers use AI as a sophisticated clipboard. Here's what changes when you build the infrastructure to make it an actual collaborator.
Read more →What does an AI model's internal representation of your personal memory actually look like? We ran 484 documents through two frozen models and mapped the geometry. Here's what we found.
Read more →We scaled the geometry mapping experiment to 2,000 documents and three models. The results changed in ways we didn't expect — and pointed directly at what Aislinge needs to do first.
Read more →Understanding the landscape of AI models — from transformers to diffusion models — with practical applications, tradeoffs, and what's changing in 2025.
Read more →A comprehensive guide to prompt injection vulnerabilities in LLM applications, from basic attacks to advanced defense strategies.
Read more →