Phil Whittaker

MAR 3

AI models have crossed a genuine capability threshold. The question is no longer whether they can write production code—it's whether you're willing to believe it. Here's how to make that shift.

FEB 24

Skills and MCPs solve fundamentally different problems - one excels at information delivery and context management, the other provides structured tool integration. Here's why both matter.

FEB 4

An introduction to Umbraco CMS Backoffice Agent Skills - AI-powered skills that give coding assistants structured, up-to-date knowledge for building Umbraco backoffice extensions correctly.

JAN 11

A cost and performance analysis of different Claude LLM models for the Umbraco CMS Developer MCP, revealing that smaller models with well-structured prompts can outperform larger ones.

NOV 3

Exploring why AI is fundamentally meta - it transforms your input into structured output layer by layer. Understanding this principle unlocks more effective AI collaboration.

AUG 7

How AI-assisted coding mirrors the creative direction of Damien Hirst and Kanye West - great work comes from vision, curation, and collaboration, not just individual execution.

AUG 5

AI-assisted development as a tennis match - strategies for serving strong prompts, managing multi-turn rallies, and staying in control with Claude Code's advanced features.

JUL 30

Why playful experimentation with AI tools is the key to thriving as a developer - lessons from a BNY Mellon AI meetup and a lifetime of curiosity-driven learning.

JUL 20

A step-by-step guide to building a headless Umbraco site using the Clean Starter Kit with a Next.js frontend, covering the Delivery API, Orval, revalidation, and image optimisation.

JUN 30

Going beyond API wrappers with MCP - fine-grained strategies for parameter schemas, tool descriptions, pagination, context management, and multi-server composition.