Trusted Reasoning
Working notes on where AI creates real operating value and where it quietly stalls.
Decision Intelligence
Why consequential AI needs to understand not only what is happening in the business, but what the organization has decided to do about it.
August 18, 2026
The Thesis
The uncomfortable truth behind stalled AI: the models usually work. The data underneath them doesn't.
June 28, 2026
What I’m exploring next
How accurate does enterprise AI actually need to be?
Accuracy should be judged against the decision, the consequence of being wrong, and the kinds of errors that matter most.
When does AI actually create business value?
A better prediction or recommendation is not the end point. Value depends on whether it changes a decision or action, produces an outcome, and can be measured against what would otherwise have happened.
What should AI decide and act on, and what should still require human judgment?
As AI becomes more capable, the challenge is not keeping people in every loop. It is defining where AI can act autonomously, where uncertainty or consequence requires escalation, and how accountability should work.