AI has changed how quickly I can build, but I do not want speed to replace engineering decisions. I use AI heavily for implementation, investigation, refactoring, and repetitive work. The important part is keeping the decisions visible. I still want to know why a table exists, why a route is structured a certain way, and what trade-off a library introduces. My preferred workflow is simple: define the problem, agree on the architecture, implement in small steps, review the result, and verify the behavior.
Building With AI Without Skipping Engineering
How I use AI-assisted development while keeping architecture, review, and product decisions explicit.
