Every Mostly Right issue, kept together. Open a letter in your browser or download the PDF to read later.
Organizational capability / September 22, 2026
Eight in ten workers say AI made them personally more productive this year, per McKinsey's 2026 State of AI survey of 1,719 executives across 97 nations, but only 6% of companies can attribute a real earnings impact to it.
Manager enablement / September 17, 2026
Only 28% of employees say their manager actively supports the team’s AI use, according to a November 2025 Gallup survey of 19,043 U.S. employees, and that gap predicts almost everything else about whether AI adoption sticks.
Enablement / September 14, 2026
AI training should pair shared fundamentals with practice built for the work employees, managers, and leaders actually do.
AI at work / September 8, 2026
A useful workflow can save two hours. That does not mean the business has changed how work gets done.
Measuring AI at work / September 1, 2026
Generation speed is only useful when trust is cheap. Measure the whole task, including the work required to trust the result.
Work design / August 28, 2026
AI may be able to do the work. The harder question is who should hold the authority—and the consequences.
Workplace culture / August 25, 2026
A Slack agent that decides when to speak changes more than coordination.
Memory & personalization / August 17, 2026
Why an AI can remember the truth and still give advice for a life that has changed.
Decision-making / August 13, 2026
AI shopping assistants may change more than checkout speed. The deeper shift is who businesses have to persuade before a choice is made.
Work design / August 10, 2026
When work says “use AI,” choose one small job where you can still judge the result.
Decision quality / August 6, 2026
In one reported two-run test, a letter explained the family decision behind an eight-month gap. The advice got more useful; its core diagnosis held.
Work design / August 3, 2026
A recommendation can keep its direction while the conditions that support it remain unanswered.
Memory & decision-making / July 30, 2026
Memory can make advice fit your life. A review process helps it keep fitting when your life changes.
Personal AI / July 27, 2026
An eight-question interview gave Don Long something more powerful than a better prompt: a system that could use his own history to argue against him.
Decision quality / July 23, 2026
A campaign’s working estimate beat its target. Six conversations with Claude recommended holding or deferring. The useful question was what each answer had assumed.
Decision quality / July 20, 2026
A new conversation can show whether a recommendation depends on the story that came before it. A useful check, not a verdict.
Decision quality / July 13, 2026
A mortgage choice and a 401(k) question surface the same gap: an answer can sound settled before the system has learned which facts could change it.
