Frameworks

The tools you’ll keep using on Monday.

These are durable, tool-agnostic frameworks: they work regardless of which AI model you’re using or which one comes out next quarter.

The trust tool: when to trust AI

The core lens for the whole day: Average / Precise × Small / Large.

  • AI excels at average answers over small, low-stakes work, so lean in.
  • It gets dangerous on precise answers and large, high-stakes work, so keep a human in the loop.

This is what makes “where does a human stay in the loop?” an answerable question rather than a vibe. It comes from the free companion book, Conversation, Not Delegation. Start there to go deeper.

The trust tool: a 2×2 of how exact the answer must be (average → precise) against stakes and scope (small → large).

DDCD: leading through the inevitable crisis

When an AI project hits trouble (and it will), work the problem: Diagnose → Decide → Communicate → Document. First ask whether it’s a technical, people, leadership, or ethical crisis, as they need different responses.

Scale / Pivot / Kill: the go/no-go call

At each gate, the honest question isn’t “is it perfect?”. It’s whether the evidence justifies continuing. Killing a project that isn’t working, early, is success, not failure. The cost of wrongly scaling is far higher than the cost of wrongly killing.

Take-home references