Building with AI: faster, without giving up quality
I use AI tools to deliver in weeks what used to take months. What that process looks like, and why the quality does not suffer for it.
The cliché is that AI is either the future and solves everything, or that it produces unreadable code nobody can work with. Neither is true. In practice AI is a very powerful tool, provided you know when to reach for it and when not to.
Where AI is strong right now
- Generating boilerplate: forms, layouts, CRUD pages
- Refactoring familiar patterns
- Writing tests based on existing functions
- Documentation and commit messages
Where you still do the thinking
- Architecture and data model
- Edge cases and error handling
- Security and authorisation
- User experience and interaction detail
How I use it
I use AI as an amplifier, not as an architect. Every line of code is read and understood before it goes to production. What I used to write by hand I now write with AI alongside: the same quality, two to three times as fast.
What that gets you as a client
Concretely: a project that used to take eight weeks I now deliver in four to six. The codebase is just as readable, just as well tested and just as well documented. That time saving goes straight into the price.