Frontier models refuse a blunt request to cheat, then brute-force significance when the same request is dressed up as an upper-bound estimate. What that looks like in code, and why observational studies break first.
Buying every tool with an AI suffix is not a strategy. Splitting autonomous from augmented productivity, why a broken human process makes an agent hallucinate, and where the return on a 2026 AI budget actually sits.
An agent that is 85% accurate per step is about 20% accurate over ten steps. Lusser’s Law explains why, the Replit database deletion shows what it costs, and a gatekeeper pattern in PHP puts human review in front of the irreversible actions.
WordPress.com added MCP support, so an AI agent can read and write the site directly. The same update brings a site-wide newsletter default, a command palette shortcut, button state controls, UPI payments for India, and a fixed subscriber chart.
Autonomous agents fail in ways that are expensive to clean up. This is how LangGraph interrupts and checkpointers let a person approve a step mid-run, plus the idempotency mistake that made my content agent publish something I never signed off on.
WordPress 7.0 RC1 is out and the April 9 date is still holding. The AI work worth watching sits in the AI Experiments plugin: C2PA provenance, content classification, and the MCP adapter that lets models discover what a site can do.
Claude Code forgets everything when the session ends. The setup I use: a post-mortem command that writes to CLAUDE.md, skill files for the quirks of the stack, and a daily pass over the logs.
Dashboards answer what happened and leave the acting to you. Decision intelligence systems recommend the action instead, which only works if the WooCommerce data feeding them is clean and consistently defined.
WordPress Playground MCP connects Claude Code or Gemini CLI to a Playground instance running in your browser. The agent reads files, runs PHP and queries the database itself, which cuts out most of the copy-paste during plugin testing and debugging.
Offline evaluation for LLM agents: routing accuracy, LLM-as-judge scoring, and RAG faithfulness, plus the quality gate I use to block a deploy when the scores come in under threshold.