For the past two years, most of my code has been written together with AI. The cost of taking a feature from idea to production has dropped faster than at any time in my career. I expected to be excited, and at first I was. But over time I noticed something strange: I was shipping faster, yet customers were not more likely to stay. In this article I want to share this observation, and how ITIL 4 helped me think it through.
Software’s old moat was the ability to build. Skilled builders were rare and building was slow, so “we can build it” was an advantage by itself. AI has flattened most of that wall. When every team can produce something functionally adequate, differentiation cannot come from features anymore. In other words, scarcity did not disappear; it moved, from development to delivery.
ITIL 4 offers clean language for this. Strictly speaking, it uses Utility and Warranty to describe whether a service is fit for purpose and fit for use, while Drive Stakeholder Value gives customer and user experience explicit attention. I place Experience beside the other two as a practical review lens, not as an official fixed triad. AI has lowered the time and cost of some functional development, but not to zero, and it does not give you Warranty or Experience for free. Ownership at 3 a.m., the first hour of a breach, and a clean handoff from automation to a person all require operating decisions for which the organization remains accountable.
This matches what my long-time partner Yu-kai Chou describes in his Digital Convergence Model: technology, behavior, and business models are converging, and as the technology side converges, competition moves to behavior and experience. Everything I have seen in SaaS and B2C says the same: feature lists get customers to try you; experience and delivery quality decide whether they stay.
This is why service management matters in the AI era. In practice it means how fast the right person takes over an incident, whether recurring causes are removed, whether SLAs are managed as promises, and whether each critical touchpoint has an owner. AI can assist with routing, summarization, and automation; it cannot assign accountability on the organization’s behalf. The meaningful gap is no longer only shipping speed, but whether the service can be carried after release.
That is why I completed the ITIL 4 Master path (the full journey is in “From Developer to ITIL 4 Master”). The next time you review an AI feature, ask three questions: who takes over when it fails, how the service promise is measured, and how a stuck customer reaches help with context. Those answers reveal delivery capability more clearly than release speed.
Sources and version notes
- PeopleCert: ITIL 4 Drive Stakeholder Value and customer/user experience
- PeopleCert: ITIL Experience (Version 5)
- Yu-kai Chou: Digital Convergence Model