AI coding assistants and the WordPress code they get wrong
AI coding assistants write PHP that reads fine and falls over in production. A WooCommerce audit, the naive fetch they keep producing, and the refactor I run it through.
Category: AI in WordPress
Gemini Embeddings 2: one vector space for text, images and video
Gemini Embeddings 2 is in public preview: one vector space for text, images, audio and video. The preview limits, calling it from WordPress with the HTTP API, and why cosine similarity on its own still is not enough.
Prompt caching: stop paying twice for the same tokens
Prompt caching takes up to 90% off input token cost and around 80% off latency, but only when the prompt prefix never moves. Here is how the pre-fill stage works, where the 1,024-token threshold sits, and how to order a prompt so it hits the cache.
Building an AI application is mostly infrastructure work
Building an AI application is mostly plumbing: keys kept out of the repo, a plan for 429 responses, and a chunking engine for anything longer than the token limit. Notes on what keeps an AI feature running once it leaves your laptop.
Vision language models are not trained from scratch
Nobody trains a vision language model from a blank slate. You freeze a ViT, train a Q-Former to bridge image features into the text embedding space, and wrap the language model in LoRA. This is how those three pieces fit and where the GPU hours go.
Causal inference methods that beat your dashboard
A dashboard shows you what moved. These five methods tell you what caused it: doubly robust estimation, instrumental variables, regression discontinuity, difference-in-differences, and CATE, with the Python I run on live WooCommerce data.
Why your data stack is not ready for AI agents
Dashboards are no longer the main consumer of your data. Here is what I would change in a data stack to make it readable by agents: fewer tools, open table formats, a real semantic layer, and feedback loops that act without waiting for a human.
Why human strategy still runs an AI-accelerated workflow
AI raised the floor on production, and that changed what designers are actually paid for. Here is where human strategy still decides the outcome, and why judgment now carries the accountability.
Machine learning at scale: availability beats accuracy
Machine learning at scale is an infrastructure problem before it is a modeling one: why availability beats accuracy, how label leakage fakes good metrics, and why heavy models and fallbacks belong on different hardware.
How I get production-ready code out of Claude Code
Vibe coding a feature is easy; keeping it out of the technical debt pile is not. A CLAUDE.md handbook, plan mode before any edit, and a review pass that assumes something is broken.