Everything You Learned About Building Software Is Already Wrong
You read Brooks. You read Fowler. You read The Phoenix Project and highlighted the good parts. Then you built something on Claude Code this weekend and realized your entire engineering organization is obsolete. Here is what building software actually looks…
The shift to agent-driven development changes the fundamental economics of software production, moving the constraint from human capacity to the capacity for human judgment.
Treat AI as an inflection, not an iteration.
The conventional wisdom of software development emerged from a constraint set where human labor was the primary factor of production; this constraint no longer holds with agent-driven systems.
Value streams optimized for human-scale iteration become liabilities in an agent-driven paradigm, increasing lead time and overhead rather than ensuring quality or coordination.
Capability, defined as the ability to effectively direct and evaluate agentic output, becomes the core differentiator and primary investment area for organizations adopting AI in their SDLC.
The cost arbitrage of distributed human labor diminishes as a single highly capable individual, augmented by agents, can achieve the output of a much larger, less-augmented team.
The first question for any AI program: are we optimizing the legacy process, or are we designing for the new constraint?
The shift to agent-driven development changes the fundamental economics of software production, moving the constraint from human capacity to the capacity for human judgment.
Treat AI as an inflection, not an iteration.
The conventional wisdom of software development emerged from a constraint set where human labor was the primary factor of production; this constraint no longer holds with agent-driven systems.
Value streams optimized for human-scale iteration become liabilities in an agent-driven paradigm, increasing lead time and overhead rather than ensuring quality or coordination.
Capability, defined as the ability to effectively direct and evaluate agentic output, becomes the core differentiator and primary investment area for organizations adopting AI in their SDLC.
The cost arbitrage of distributed human labor diminishes as a single highly capable individual, augmented by agents, can achieve the output of a much larger, less-augmented team.
The first question for any AI program: are we optimizing the legacy process, or are we designing for the new constraint?
After 20 years in software development, Norman is both a hands-on leader and defining the new age of AI SDLC for some of the biggest brands in the world — and exploring it with the builders. He writes here about things he is hearing and seeing. All posts are his personal points of view and do not reflect any employer or any customer he has ever had contact with.
The views and opinions expressed in this article are the author’s own and do not represent the positions of any employer, client, or affiliated organization.
