A field report from the frontlines of AI-assisted development at scale — why language models optimize locally, how that corrupts architecture over time, and what I am doing about it.
The Innovator's Dilemma framework applied to the AI revolution in SaaS. When development costs drop 90%, the playbook for disruption becomes uncomfortably clear.
Building software is maybe 10% of the job. The other 90% is living with what you built. Senior developers understand this—and it changes how they write every line of code.
Model Context Protocol promises seamless AI tool integration, but the reality of shipping it reveals quality issues, configuration hell, and premature standardization that creates more problems than it solves.