AI in Practice

Working with AI systems directly — models, tools, workflows and the knowledge infrastructure behind them. From hands-on experiments with Claude, ChatGPT and Gemini to structuring content for intelligent systems.

AI in Practice

GPT Image 2 Changed My Mind on AI Visuals

I had quietly written ChatGPT off for image generation. Then GPT Image 2 showed me a technical diagram of a washing machine, and I had to revise that judgment.

AI in Practice

Dressing up an AI Model in a Harness

An engineer friend calls the infrastructure around an AI model a harness. The word is technically accurate and instinctively wrong. What that tension reveals.

AI in Practice

Built in, not bolted on

Most organisations are adding AI to software that was never designed for it. The real gains come when AI is part of the architecture, not an add-on to it.

AI in Practice

The Workspace You Already Have

Knowledge workers have built informal AI workspaces from Claude, Gemini, and Google Drive. The gap isn't the tools. It's governance, sharing, and where the data lives.

AI Governance

When Git Grows Up

Git-based knowledge management has always been too complex for content teams. Agents change that, and the implications for governance-heavy organisations are worth watching.

AI in Practice

Delegating Past Your Own Ceiling

Cloudflare is powerful enough to extend anything Ghost can't do. The problem is complexity. Claude Code changed that by operating the layer I previously couldn't reach.

Guests That Should Behave

Bots are modern guests. Most are welcome. But when they arrive in disguise with a real browser and spike your analytics, hospitality has limits. A Cloudflare story.

AI in Practice

Claude Code Routines and the Access Problem

Claude Code's new Routines feature ran my Google Calendar briefing immediately, no setup. The reason: authentication was already there.

AI in Practice

The Claude plan I actually use

Pro subscription, a prepaid credit buffer, and Sonnet for most things. For a non-coder working with AI daily, this turns out cheaper than the Max plans.

AI in Practice

The Neural Engine Does Not Run Your LLM

My earlier post on AI hardware implied the Neural Engine handles on-device AI. It doesn't, not for LLMs. Here is what it actually does, and why the distinction matters.

AI in Practice

Running Gemma 4 on Your iPhone

Google's Gemma 4 runs offline on your iPhone. A follow-up to my local LLM experiment, now with a sharper app, a better model, and a clearer sense of what this category is becoming.

AI in Practice

My First Mac App

I described a workflow problem to Claude. It wrote me a compiled Swift app using Xcode. That sentence still feels odd to type.

AI in Practice

ChatGPT Thinks With You. Claude Builds With You.

I switched from ChatGPT to Claude. The difference isn't the model. It's the shift from cognitive support to digital crafting, and what that makes possible.

AI in Practice

Back to the file

Thirty years of web technology, and the most reliable setup is still a text file on a server. We just needed AI to get back there.

Future of Work

A lighter computer for a different era

A new iMac, a one-hour migration, fewer ports, more cloud and AI. Replacing a desktop computer now reveals how much the role of the machine itself has changed.

Future of Work

The Delegation Problem of the Internet

Modern life runs on delegation. Yet most digital systems assume one account, one person, one operator. Passwords and identity checks quietly block automation.

AI in Practice

A Personal Retention Strategy for ChatGPT

From digital scarcity to data abundance, and why I now choose deliberate reset over permanent retention in ChatGPT.

AI in Practice

What coding with AI feels like now

There is a lot of noise about AI replacing programmers. I wanted something else: a sense of what is feasible, what is throwaway, and where judgement still matters.

AI in Practice

ChatGPT Memory Has Quietly Changed (Plus & Pro)

If you use ChatGPT Plus or Pro: memory now goes beyond saved preferences. Past conversations can be retrieved when relevant. Business accounts are different.

AI in Practice

ChatGPT Personal vs Business

If you use ChatGPT both privately and at work, you may have noticed this: one conversation seems to carry on over time, the other keeps resetting. Same system, different behaviour. That difference is deliberate, and it matters.

Word of the year: model

The word model comes from modulus, Latin for ‘measure’. It refers to simplified representations we use to make sense of reality, from maps to AI systems.

Why AI Image Editing Is Often Rebuilding

AI can make true pixel edits, but many image “fixes” work by reconstruction. Typography and colourisation expose where rebuilding replaces editing.

Structuring Knowledge with Obsidian

Discovering how Obsidian, Markdown and a light taxonomy can reveal the hidden structure of five years of writing and open a path toward knowledge engineering.

AI in Practice

Making Content Ready for Intelligence

Unstructured content already holds the knowledge organisations rely on. The shift is learning how to reveal its structure so intelligent systems can use it.

AI in Practice

Voice Mode Without the Screen

Voice mode with Gemini and ChatGPT. On iOS the lock screen, Dynamic Island, and Live Activities decide whether your assistant stays visible.