Using AI Tools in My WordPress Development Workflow: What Works and What Doesn't
I've spent the last six months integrating AI coding tools into my day-to-day WordPress development. Here's an honest account of where they save real time and where they still fall short.
Topics covered
- AI coding assistants
- Cursor & GitHub Copilot
- WordPress-specific limitations
- Practical workflow integration
The Context
I was sceptical of AI coding tools for most of 2024. The output I saw early on was often confidently wrong — particularly for WordPress-specific patterns, where training data is mixed with outdated tutorials and deprecated approaches. By late 2025, the tools had improved significantly, and I started integrating them seriously into my workflow. Six months on, they've changed how I work — but not in the ways I expected.
Where AI Assistants Genuinely Help
Boilerplate generation is where AI tools save the most time in WordPress development. Registering a custom post type, setting up a WP_Query with a complex set of arguments, writing a REST API endpoint with proper authentication checks — these are patterns I know well but still take time to type correctly. Having an AI complete them from a brief comment or function signature is a genuine time saver, not because I couldn't write the code, but because I don't have to. The cognitive load of holding the whole task in mind while typing boilerplate is reduced, which leaves more focus for the parts that actually require thinking.
Writing and Fixing Tests
I've started writing more unit tests for WordPress projects — partly because AI tools make the test-writing step less painful. Describing what a function should do and having the assistant generate a PHPUnit test case is faster than writing it from scratch, and the output is usually good enough to be a useful starting point. Where AI also helps is debugging failing tests: pasting the failing test, the function under test, and the error message into the chat interface and asking for analysis has surfaced real bugs quickly, rather than requiring me to mentally trace through the code myself.
Where It Still Falls Short
WordPress-specific context is the biggest limitation. AI tools have good general PHP knowledge but often suggest patterns that are technically correct PHP but wrong for WordPress — using PDO directly instead of $wpdb, not checking nonces on form submissions, or suggesting approaches that ignore WordPress's hook system. On complex projects with custom database tables, bespoke APIs, or intricate hook interactions, the AI's suggestions need heavy review. It's not a replacement for WordPress expertise; it's a tool that's most useful when you're experienced enough to evaluate its output critically.
Cursor vs GitHub Copilot
I've used both extensively. Copilot integrates cleanly into VS Code and is unobtrusive — suggestions appear inline, and I accept or reject them without breaking flow. Cursor's chat interface is more powerful for multi-step reasoning and debugging, but switching between editor and chat adds friction. My current setup uses Copilot for completion during active coding and a separate AI chat interface for debugging, architecture questions, and anything that requires extended context. Using both for their respective strengths outperforms either alone.
My Honest Assessment
AI coding tools have made me faster on the parts of WordPress development that are repetitive — not on the parts that are genuinely hard. Designing a clean architecture for a complex plugin, debugging a subtle race condition in an AJAX handler, or figuring out why a WP_Query isn't returning what you expect — these still require deep WordPress knowledge and careful thinking. The tools haven't changed that. What they've done is compress the time I spend on everything else, which gives me more time and cognitive bandwidth for the problems that matter.
Written by Manan Vyas
Senior WordPress Developer · Manchester, UK