QFM103: Engineering Leadership Reading List - February 2026

This piece examines StrongDM's "Software Factory" approach, where non-interactive AI agents write code driven by specs and scenario-based tests, with no human code review. The author explores the core challenge of verifying agent-produced software when both implementation and tests are generated by AI, and notes StrongDM's benchmark that teams should be spending at least $1,000 per engineer per day on tokens if the factory is operating at full capacity.

The author describes their practice of keeping a physical engineering notebook, started during consulting work, where they write dated, append-only notes about their work -- including sketching out code changes by hand before typing them into an editor. They argue the notebook serves as both a memory aid (handwriting being more effective for retention) and a tool for thought that forces deliberate reasoning before implementation.

The article argues that the hardest part of maintaining focus as a leader is not rejecting bad ideas but learning to turn down genuinely good ones, because saying yes to everything dilutes team capacity and leaves existing work incomplete. It offers a framework for evaluating incoming ideas: asking whether they contain new information that changes your assumptions, and then deciding whether the right response is a pivot, a "not now," or a firm no.