About WhenMachines

The short version

WhenMachines explains AI by looking at mechanisms before marketing. We want to know what a system does, what evidence supports that conclusion, where it breaks, and what changes when it reaches real users.

That means being anti-hype in both directions. We do not treat a launch deck as proof, and we do not mistake a model's limitations for proof that nothing useful is happening. The interesting answer is usually more specific than either story.

What makes this site different

We do not regurgitate press releases, rewrite vendor claims, or turn a benchmark score into a verdict. A useful explanation starts with the primary material: documentation, model cards, papers, release notes, and hands-on testing when it is possible. Then it separates demonstrated behavior from sales language.

The goal is not to make AI sound inevitable or ridiculous. It is to give you a clean read on how a tool works and whether it deserves your attention.