Latest Blog Posts

ChatGPT's Privacy Mess: Same Story, Different Day

Google has crawled over 70,000 ChatGPT conversations that people assumed were private. Search "site:chatgpt.com/share" and you'll find resumes, company secrets, therapy-like conversations, and personal drama, all sitting there for anyone to read.

How I built an AI assistant that analyses my Linear tickets every morning, prioritises my work, and sends me a personalised daily plan via email.

Automating my morning with Linear MCP Server and Claude

Every morning, I'd sit down with a cup of Yorkshire tea (only the best) and do the same routine. I decided to automate this entire process using Claude AI and Linear's new MCP (Model Context Protocol) server. Now, every weekday at 7 AM, I get an email with a plan of how to take on the day.

How I used Claude and a systematic template approach to generate 1,100+ hours of writing in just 3 days

From template to 138 published articles: How I built an AI-powered content engine that generated 640K words in 3 days using Claude and automation.

How I automated the boring parts of content creation while keeping the creative control

Building a Content Workflow Tool with Claude AI

As someone who writes a lot of different things from technical docs to blog post to marketing copy I was increasingly noticing there were some repetitive parts of content creation. Not the creative thinking or strategic decisions, that stuff is still very much required but the more mechanical stuff: adapting tone for different audiences, ensuring consistent company messaging, and doing the same types of edits over and over.

Researchers from Apple claim these models face "complete accuracy collapse" when problems get complex enough. Critics, including researchers and models from Anthropic (yes their model Opus is one of the authors), argue these failures are a result of flawed experiments, not flawed models.

Do Reasoning Models Really Think? The Unlikely Apple-Anthropic Research Feud

A recent academic dispute brings questions if reasoning models actually exhibit genuine reasoning, or just sophisticated pattern matching dressed up as thought? Researchers from Apple claim these models face "complete accuracy collapse" when problems get complex enough. Critics, including researchers and models from Anthropic (yes their model Opus is one of the authors), argue these failures are a result of flawed experiments, not flawed models.