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.
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.
Peeking at results, skipping the power calculation and tracking twenty metrics turn most A/B tests into noise. Here is the pre-test protocol I use on WordPress and WooCommerce sites, plus the short checklist I run before starting a test.
WordPress 7.0 Beta 5 adds a Command Palette shortcut to the admin bar and closes more than 101 bugs. What the shortcut does, how to register your own commands with the API, and why I update beta sites with WP-CLI.
Most like-for-like reporting gets solved in DAX and then runs slowly. This is the model I build instead: a bridge table with a store-month key, the comparison logic in SQL and Power Query, and a sales measure that is nothing but SUM.
Moment.js is in maintenance mode and still costs you bundle size. The JS Temporal API is native, immutable and Stage 4, so this walks through creating dates, arithmetic, Intl formatting and a time zone refactor in a WooCommerce extension.
WordPress Studio added Xdebug support and a dedicated debug log toggle. Here is how I turn both on, the one-site-at-a-time catch, and the port your IDE has to listen on.
PythoC compiles a subset of Python into standalone native binaries through LLVM. My notes on the 40x Fibonacci run, the print() that produces nothing, and the cases where a native compiler is the wrong tool.
Real test data is usually missing right when you need it. How I generate varied synthetic WooCommerce orders with an LLM, and the two ways it goes wrong: context rot, and only ever prompting for the happy path.
Pure semantic search stops finding part numbers and SKUs. Why I pair embeddings with a BM25 keyword index, what the k1 and b parameters change, and how rank fusion merges the two result sets.