LLM encourages everyone to take a shortcut. The Software Architect uses LLM to generate specs, introducing hallucinations. The developer uses LLM the generate the code, introducing errors. The review uses LLM to review the code, also propagating the issue. When the code fails in the hand of real users, nobody knows how to fix it. Nobody bothered to learn what the code was actually supposed to do.

Join me in this episode as we tackle the problem and figure out a solution for the developer workflow.