Mental Models for Recurring Decisions

The Platform vs. Product Test: When a team requests a new capability, ask: is this something multiple teams will need? If yes, build it as a platform capability. If no, let that team build it. Only centralize when duplication cost exceeds coordination cost.

The Experiment Budget Test: What is the maximum we can learn for the minimum we can spend? Define the hypothesis in falsifiable form. Define the minimum experiment that could invalidate it. Fund that experiment, not the full initiative.

The Symbiosis Audit: Where is the human in this loop, and are they in the right place? If the human is doing what a machine could do faster, automate it. If the machine is making a decision that affects human lives without oversight, redesign it.

The Velocity Diagnostic: When a project is moving slowly, ask: where is the waiting state? Waiting for approval? Replace approval with guardrails. Waiting for another team? Build an API. Waiting for a meeting? Make the decision asynchronous.

The Ethics Red Team: Before shipping any intelligent system, convene a diverse group and ask: how could this be used to cause harm? Test for demographic bias. Design for transparency. Build human override mechanisms.