Developing in an Ever-Changing AI Climate
AI is narrowing the distance between ideas and working solutions—and making clarity, architecture, judgment, and user-centered design more valuable than ever.
Read full recordWorking notes
Ideas for making systems more useful, dependable, and human.
Notes from the intersection of operations, product thinking, software, and AI-assisted building.
Operational knowledge belongs in the architecture
The rules people carry in their heads are already part of the system. The only question is whether the design acknowledges them.
A process diagram can look clean while the real work depends on exceptions, timing, judgment, and a dozen unwritten handoffs. That hidden layer is not noise around the system. It is the system doing its job.
Good architecture makes that knowledge explicit without flattening it. It gives ordinary cases a clear path, protects room for judgment, and records why an exception exists. The result is more than documentation: it is a tool that respects the people who understand the work.
AI-assisted development without surrendering the design
Speed is valuable only when you remain responsible for the shape of the system.
AI can shorten the distance between an operational insight and a working interface. It can draft, compare, test, and expose paths that would have taken much longer to explore alone. That leverage is real.
But generated code is not a product decision. The builder still owns the model, the security boundaries, the failure states, and the standard for what belongs. I use AI as a capable collaborator while keeping the system’s intent legible enough to question and maintain.
Build for the person doing the work
The best interface is not the one that displays the most system. It is the one that removes the right uncertainty.
People rarely arrive at a tool hoping to admire its information architecture. They arrive with a task, incomplete context, and a consequence if something goes wrong. Design starts by understanding that moment at full resolution.
That means naming things in the language of the work, placing decisions in sequence, making state visible, and treating recovery as a first-class path. A dependable tool quietly returns attention to the work itself.