<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Andy Max&apos;s Web Log</title><description>A small personal web log about code, infrastructure, AI, old internet culture and whatever else gets stuck in my head.</description><link>https://andymax.co.uk/</link><language>en-gb</language><item><title>What happens when you watch all the UK news channels at once?</title><link>https://andymax.co.uk/weblog/broadcast-prism/</link><guid isPermaLink="true">https://andymax.co.uk/weblog/broadcast-prism/</guid><description>Six weeks of recording, transcribing and comparing UK television news across eight channels — and what the data shows so far.</description><pubDate>Mon, 10 Aug 2026 00:00:00 GMT</pubDate><category>media</category><category>ai</category><category>data</category><category>broadcasting</category></item><item><title>The Homelab Holiday Heuristic</title><link>https://andymax.co.uk/weblog/homelab-holiday-heuristic/</link><guid isPermaLink="true">https://andymax.co.uk/weblog/homelab-holiday-heuristic/</guid><description>A real homelab outage during a family holiday, and what happened when an AI agent was given SSH access to diagnose and repair it remotely.</description><pubDate>Wed, 29 Jul 2026 00:00:00 GMT</pubDate><category>homelab</category><category>ai</category><category>infrastructure</category><category>incident-response</category></item><item><title>A cat flap, two air ducts, and an AI that does CAD</title><link>https://andymax.co.uk/weblog/cat-flap-ai-cad/</link><guid isPermaLink="true">https://andymax.co.uk/weblog/cat-flap-ai-cad/</guid><description>How I used AI to design and print a bespoke duct transition part for a cat flap hole — without opening a CAD program.</description><pubDate>Mon, 22 Jun 2026 00:00:00 GMT</pubDate><category>ai</category><category>3d-printing</category><category>cad</category><category>diy</category></item><item><title>AI agents are for the things you keep not doing</title><link>https://andymax.co.uk/weblog/shield-debloat/</link><guid isPermaLink="true">https://andymax.co.uk/weblog/shield-debloat/</guid><description>After years of meaning to debloat my Nvidia Shield, I used Hermes to do it in 20 minutes. Not because I couldn&apos;t, but because I never would.</description><pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate><category>homelab</category><category>android-tv</category><category>hermes</category></item><item><title>Two modes for your coding agent: daily and hard</title><link>https://andymax.co.uk/weblog/two-modes-coding-agent/</link><guid isPermaLink="true">https://andymax.co.uk/weblog/two-modes-coding-agent/</guid><description>Stop running one coding agent config for everything. Set up a daily driver and a hard mode, and switch when the task actually needs it.</description><pubDate>Tue, 09 Jun 2026 00:00:00 GMT</pubDate><category>ai</category><category>coding-agents</category><category>opencode</category><category>dev-tools</category></item><item><title>What I Actually Use an AI Agent For</title><link>https://andymax.co.uk/weblog/what-i-use-ai-agents-for/</link><guid isPermaLink="true">https://andymax.co.uk/weblog/what-i-use-ai-agents-for/</guid><description>How I use Hermes Agent as a practical operator in my homelab — school meals, server health, coding, research, and the boring jobs that quietly save time.</description><pubDate>Mon, 18 May 2026 00:00:00 GMT</pubDate><category>ai</category><category>homelab</category><category>agents</category><category>automation</category></item><item><title>Do Not Underestimate the Pineapple</title><link>https://andymax.co.uk/weblog/do-not-underestimate-the-pineapple/</link><guid isPermaLink="true">https://andymax.co.uk/weblog/do-not-underestimate-the-pineapple/</guid><description>I asked several local AI models to draw an ASCII pineapple. Many failed in odd ways. Here is what went wrong, and what it says about how models think about text and space.</description><pubDate>Thu, 14 May 2026 00:00:00 GMT</pubDate><category>ai</category><category>llm</category><category>ascii-art</category><category>prompting</category></item><item><title>The Tetris Test: Comparing GLM 5.1, Kimi 2.6, DeepSeek, Gemma and Sonnet</title><link>https://andymax.co.uk/weblog/the-tetris-test/</link><guid isPermaLink="true">https://andymax.co.uk/weblog/the-tetris-test/</guid><description>Testing alternative coding models on the same Tetris clone task — GLM 5.1 won, Kimi 2.6 was close, and the results were surprising.</description><pubDate>Wed, 13 May 2026 00:00:00 GMT</pubDate><category>ai</category><category>coding</category><category>llm</category><category>testing</category></item></channel></rss>