skip to content
Terry Li

I spent a morning building a compound machine — a weekly pipeline that takes AI news and produces consulting IP: policy templates, reference architectures, use case libraries. Flywheel logic. Each week’s output feeds the next.

Then I tried to apply the same pattern to the rest of my life. Exercise. Marriage. Financial planning. And hit a wall.

Not every goal is a flywheel. Some are checklists — execute a will, migrate an investment account, buy term life insurance. You do them once and they’re done. Building a “financial resilience system” around them would be absurd.

Some goals are habits. Going to the gym doesn’t compound from better planning. It compounds from going. The system is the enemy of the doing.

And some goals are pure attention. Being present with a toddler. Protecting time with your partner. No agent, no pipeline, no dashboard helps. The only input is you, undistracted.

The mistake isn’t building systems. It’s applying the flywheel pattern to everything because that’s the tool you know. When you’re holding a compound machine, every goal looks like it needs compounding.

The diagnostic is simple: does the output of this goal feed back as input? Career knowledge does — each insight enables the next. Financial tasks don’t — the will is done once. Gym visits don’t feed each other. Quality time doesn’t scale.

Match the tool to the shape:

  • Flywheel → build the machine
  • Checklist → sprint and stop
  • Habit → remove friction, then discipline
  • Attention → protect time, stop building

The hardest part isn’t building the machine. It’s recognising when the machine is the wrong tool.

Related: north-star-shapes | autonomous-flywheel | automation-spectrum