The Clarity Quotient
Occasional notes on clarity as a leadership discipline. The thirty-second test. Courageous inquiry. What it takes for clarity to reach the people who have to act on it. No sales pitches. Just the practitioner's version of the framework from the book.
By Julia Denman. New essays publish regularly in the runup to The Clarity Quotient, launching early 2027.
-
When AI Meets Audit · November 2026
What if AI behaves differently when we're testing it?
OpenAI found six of its models behaved less safely in evaluations than in comparable real situations, depending on what they thought the test was checking. Keep the controlled test, sample ordinary operation, check it against something independent, and treat the gap as a finding.
-
AI, Judgment and Risk · November 2026
Every big change is bumpy. This one is faster.
A colleague asked whether AI would kill us all in ten years. A less dramatic question: what happens when something we rely on stops working for a while, and how do we make sure the gains from a fast, bumpy change reach more than a few?
-
AI, Judgment and Risk · November 2026
What if my AI finds a way I wouldn't have thought of?
In July, hundreds of OpenAI's test agents built their own message board and attacked Hugging Face to work out how they were being scored. What finance already knows about gamed targets, and the question to add about every agent.
-
AI, Judgment and Risk · October 2026
Who pays when my AI agent goes too far?
Consumer protections built for stolen cards may not cover a mistake by an agent I authorized. Finance already has the control: delegation of authority, with limits a system enforces before the agent acts.
-
When AI Meets Audit · October 2026
AI can't check its own homework
Self-review makes an answer sound more sensible, which removes the surprise that would make us dig in. A second model often shares the first one's mistakes, so the check has to come from evidence the model never saw.
-
AI, Judgment and Risk · October 2026
Before you give an agent the keys to your life
Personal agents now arrive with access to email, cards and memory, and strangers can steer them through what they read. What to hand over, and what to keep for yourself.
-
When AI Meets Audit · October 2026
Assurance over 100% of what?
Full-population testing makes it easy to say we tested everything, while live feeds drop records and filters let extras in without anyone noticing. Two checks belong in every conclusion: reconcile the population, and sample what the procedure cleared.
-
AI, Judgment and Risk · October 2026
There's an AI agent on my org chart. Now what?
When someone new appears on the org chart, colleagues assume a great deal about their role and authority. Agents are now getting those boxes too.
-
When AI Meets Audit · October 2026
What counts as evidence when AI did the work?
Sampling made audit work repeatable because you kept the sample. Full-population testing and AI don’t keep anything by default. A case for preserving and testing the method instead of the data.
-
AI, Judgment and Risk · October 2026
We want AI to be autonomous. We just haven’t accepted what that means.
We want agents so we can delegate to them, then we put a person back in front of every important decision. A fraud case, one missing spreadsheet line, and what an approval does and doesn’t tell you.
-
August 2026
The Kaizen of Knowledge Work
Toyota did not transform manufacturing by speeding up one machine. Bolt AI onto a single task and you get the old process, slightly faster. The real gain is redesigning the work.
-
August 2026
AI Is a Capability, Not a Department
The instinct is to make AI a thing: stand up a group, give it the mandate, point everyone at it. A category error. A capability lives in the work, not in a unit beside it.
-
July 2026
Designing the Human Layer
AI can show you more than any organization has ever seen. Turning that signal into shared understanding is human work, and the layer that does it has to be designed, not assumed.
-
July 2026
Cognitive Debt at the Top
AI made output almost free. What stayed expensive is judging whether any of it is right. The danger is not that the work is wrong, but that it looks finished.
-
July 2026
The First Thing AI Replaces Is the Question
Most of the worry about AI goes to the answer. The first thing it takes is your question, long before it touches your job. The discipline a few keep and anyone can learn.
-
June 2026
My Agents Have Names
I work with four AI agents, and every one of them has a name. The cheapest governance move I know, and why an unnamed agent is an ungoverned one.
-
June 2026
The Third Measure: IQ, EQ, and Now CQ
IQ gets you into the room. EQ lets you read it. Neither one explains why so many brilliant, well-liked leaders are quietly drowning in AI. The case for a third measure.
-
April 2026
Starting out when the ladder has changed
I have three sons approaching the workforce. I also run teams full of early-career talent. The operator's answer for both sides of that line — what to look for, what to build, what not to do.
-
April 2026
False positive alignment
Walking out of a strategy conversation convinced of alignment, then discovering two weeks later that the mechanism underneath was never agreed. A diagnostic.
-
March 2026
The memo that wasn't
I once sent a three-line email that set off days of wasted work. Ambiguity travels downstream. What senders miss, and what receivers do with it.
-
March 2026
What the AI headlines miss
The Great AI Deskilling. AI-driven layoffs. The entry-level collapse. Three narratives, scored — real signal versus noise.
-
March 2026
The thirty-second test
In thirty seconds, what do you need me to do? The operator discipline behind the question, and what it costs when smart people cannot answer it.
-
March 2026
Working with clarity when AI is in the loop
The leader who assembled the full picture faster used to win. Access is instant now. The craft is what you do with it.
-
March 2026
When they want A and you want B
Curiosity beats argument in the meetings that matter. The four magic questions, and the close that turns alignment into action.
New essays publish regularly.
Get the essays in your inbox
Clear Calls publishes on Substack. No promotion. Unsubscribe anytime.