Intentional, Not Reflexive: A Manager's Thoughts on AI (scottkosman.com)

Shopify CEO Tobi Lütke’s memo declaring “reflexive” AI use a baseline sparked this manager’s identity crisis and a rethink of how leaders should adopt AI. After experimenting with automation across his career, he realized the AI moment isn’t just about IC velocity but about managers using models to think better: synthesizing noisy information, framing decisions, and communicating clearly. Rather than outsourcing judgment, he uses AI for cognitive offloading—turning messy thoughts into concise summaries, refining feedback, and surfacing risks—so he can focus on coaching, listening, and strategic clarity. The piece is practical: examples include custom GPTs that ingest Jira and PR metadata to flag overloaded engineers or stalled work, a shell script driven by ChatGPT to auto-generate weekly song sheets, and a “Master Prompt” that encodes role, team, priorities, and style to get consistent outputs across tools. His recommended workflow is to start with rough notes, iterate with AI, and maintain intentionality—update the Master Prompt quarterly and know failure modes. The takeaway for the AI/ML community is that managerial value lies less in raw throughput and more in tooling that augments higher-order cognition and organizational clarity, with careful guardrails to avoid reflexive, unexamined dependence.