Essays
Notes
Shorter pieces on the practice, each tied to the frameworks.
- A Human in the Workflow Is Not a Human in the LoopOversight that never changes an outcome is not oversight. Why review turns into rubber-stamping, and what a real check needs.
- Built With Is Not Built OnMost teams calling themselves AI-native ship products that never call a model. What the label hides, and the question that settles it.
- Coordination Was Never the JobMiddle layers are being cut for the work AI does well. What survives is judgment, and the sort that shows you which part of your week is which.
- Smarter and Dumber at the Same TimeAI can make you more productive and less able to think without it. Why the early seat feels this first, and how to keep the contact.
- The Analysis You Cannot ExplainHeavy AI use without the expertise underneath shows up the moment someone asks a follow-up. Why the second question is the whole interview.
- What Can Your Agent Do Without Asking?An assistant that holds your tokens and takes actions is a privileged system. Where its approvals belong, and how far it can reach.
- The Code You Didn't AuthorizeCoding assistants arrived through the IDE, not procurement. Why suggest and act belong on different rungs, and which code stays by hand.
- The VMware MomentI once clutched my VMware licences against the cloud. The same resistance is back for coding agents, and so is the mistake on the other side.