03.How I use AI in development

I use AI assistants and agents to analyse a project’s code, prepare changes and reduce repetitive work. I use them only when I can understand what they did and check that it works.

I test them on real work to see where they lose context in the codebase. I use that to update the repository’s documentation, rules and tooling.

In practice:

1. First I check whether the change is needed. An agent can suggest rebuilding something that already exists in the project.
2. I review the result like any other contribution. I read the diff, run the tests and check that it respects the project’s boundaries.
3. I need to be able to explain and maintain every change. If I cannot, I do not use it.