AI, Tools and Transformation

Large corporations face complex software ecosystems with hundreds of applications and persistent manual tasks, making AI automation seem like a solution to eliminate software sprawl. However, the author argues this overlooks a critical reality: most workers don't think in terms of process optimization, and successful automation historically required identifying non-obvious problems and testing multiple approaches—challenges that AI code generation alone cannot solve. Simply making it easier to build tools doesn't address the fundamental difficulty of discovering what problems need solving and how to solve them.