AI Tools, Tested

This is the main line of the site. Every post is built around a real work task — I take a tool at a job I already had to do, then record what it managed, where it stalled, and whether it was worth paying for.

How I test

So that posts can be compared against each other, every test follows the same rules:

• Same task. In a head-to-head, every tool gets the identical task. I don't swap the problem halfway.
• Real projects. Tasks come from work I actually had to do, never a demo built for the post.
• Failures included. Where a tool fell over gets written up too. A review with only wins carries no information.
• Costs stated. Subscription price, token spend, and the hours it cost me.
• Dated. These tools change every few weeks, so every post carries a test date and version numbers.

The full methodology and conflict-of-interest statement are on the About page.