The demo is the easy part

Every AI demo works. Then you point the same tool at the repository you actually maintain, or the folder that has to be a report by Thursday, and it stops working. The gap between those two things is what we teach.

How we work

  • Every lesson is built on a job somebody actually has to finish.
  • Every step is on screen as it ran, not summarised afterwards.
  • Everything is worked against real constraints, not a clean room.

What we are for

We work the other way round

Most of it is filmed on a problem chosen to make the tool look good. Nothing is ambiguous, no constraint gets in the way, and it works first time. Your work is not like that.

So we take a job somebody really has and do it with the tool in front of us. When it goes wrong you see that too, because handling it is most of the skill and nobody films that part.

What is left is something you can run on your own work. It is a way of working rather than a tour of the settings, and it still holds the week the tool starts behaving differently.

The job changes from one course to the next. It might be a plugin nobody on your team wrote, or a folder of documents that has to be one table by Thursday, or something built from nothing against a deadline. The tools are different every time. The method is not.

What we make

Everything we have made, and who each one is for

Activity

The blog and the community

New pieces and new threads as they land. Both are free to read without an account.

Latest from the blog

Five ways Claude Code failed on a 240,000-line WordPress plugin

Five sessions on Easy Digital Downloads, about 240,000 lines of PHP, where Claude Code produced a confident summary and usually…

Read more

Latest from the community