Software

GenAI has permeated everyday life. The go-to answer for millions of people faced with any number of different tasks. So why does it have such a basic monochrome interface? Open a browser or app for any of the mainstream GenAIs, and you’ll be faced with some variation of a black page with a box for entering white text into. You can turn the mic on, upload a file or image, but those are ultimately context-driven sideshows. The text box is the start-point for almost all GenAI w

Under the EU AI Act (2024), AI companies are obliged to mark text generated by their LLMs as of August 2nd 2026. This means that companies who signed the EU Code of Practice on Transparency are introducing ways to do so. Failure to comply with text-marking could result in fines of up to €15 million, or 3% of worldwide annual turnover. Much coverage has been given to Claude’s announced rollout of watermarking. In reality there’s actually very little to say.

Fleet operators running electric vehicles face an asymmetry of information that costs them money every week. The vehicles generate enormous quantities of data - battery state, voltage, current draw, motor temperature, GPS, speed - every few seconds. Most of it goes nowhere. When something fails, it fails in the field: recovery time, roadside assistance, cargo delays.

Lots of rhetoric aimed at leadership teams positions the question about AI and software development as primarily whether it will replace engineers. It's the wrong question. Asking it means missing something more useful. The better question is what a single engineer can build in an hour, starting from scratch, and what that tells you about how technical work is changing.

Non-terrestrial networks are a recurring subject I've been asked about over the last year or two. Interest amongst Automotive and IoT teams has been piqued by Starlink, and high-return space stocks like RKLB and ASTS have pulled NTNs into the public consciousness. So to answer these questions I'll skip the full telco perspective here and focus on practical deployment.

The gap between a written product concept and working, tested, version-controlled code can now be closed without a developer writing a line of application code. From PRD to PR without code, automation is happening today.

Once you've played around with AI for any length of time and got under the skin of LLMs, you inevitably start to uncover the quirks, weaknesses and vulnerabilities inherent to the fundamental design concepts, and the execution itself. As seen in my other AI projects on the site, it's not hard to trigger those weaknesses once you get beyond basic GenAI and into genuine design and build scenarios.

Rubens designed the major works, established the compositional framework, and reviewed his apprentices' execution. Van Dyck eventually forked the codebase, established his own studio, and surpassed Rubens in the field he chose. Git, the version control system underpinning most of the world's software, formalises exactly this: a master repository, branches, independent lines of work, the ability to merge or to diverge permanently.

Most organisations now have an AI strategy. Very few have asked what it is doing to the planet. That is not an abstract concern. It is a capital allocation question with a measurable environmental consequence attached to it. Boards are declining to ask it. The omission will not age well.

For decades the businesses of expertise, craftsmanship and luxury have run on a simple exchange: you paid for authority because authority was grounded in the living presence of a specific person. A marketer. A strategist. A lawyer. Someone who could sit across from you, who had walked through the problem themselves, whose judgement was embodied in a living mind that you could hire and pay. Presence became the marker of truth. Not the accuracy of what they said, but the fact t

Just about every senior leader has already decided to adopt AI. The all understand the risks of not doing, and a large proportion of them understand the risks of doing so. The question some aren't asking is what happens when one of the risks becomes real, and whether they can recover from it.

The board asked for an AI strategy. What they got was a vendor presentation dressed in strategy's clothes. The vocabulary is being set by people with a commercial interest in your confusion. AI is not complicated. The industry around it is. What follows is the version of this conversation that the hyperbolic presentations skip. Here's the AI landscape for executives.

Call this the rising floor. Every useful machine absorbs the layer of work it can do better than people. The human contribution does not vanish. It moves upward, into judgement, selection, taste, and ethics, questions a machine cannot answer for you because they are not technical questions at all. That is where AI is heading now, into code, legal drafting, HR policy, marketing copy, the layer of execution that used to define a competent professional.

Evaluating Claude Sonnet 4.6 for 'reasoning redline', defensive hallucinations and structural bias.

Training a specific AI model to recognise koi and categorise them more effectively than an Anthropic off-the-shelf LLM.

Building an Anthropic-powered image recognition tool for koi breeds and individual fish

A vibe-coding horological workout for Claude, generating a fully-functional, animated Grand-Complication web application.