The Signal
One sharp idea about AI and leadership, every Tuesday - each issue with a free tool you can use on a real decision. No hype, no jargon.
Get The Signal every Tuesday
Founding readers get The Starting Line (my $47 course) free.
Issue 01The work AI cannot take from you28 Jul 2026+
The work AIcannot take from you.
The scary question is not whether AI can do your work. It is whether your calendar has been wearing a fake moustache and calling itself value.
Ninety seconds. That is how long the model needed to produce a board memo that four senior people had spent three days on last quarter, with more coffee and fewer witnesses. The room laughs at the speed, and then the laugh thins, because everyone is quietly doing the same arithmetic. Some of the work you were proud of, it turns out, was expensive formatting.
The reflex is to blame the machine, but the machine only switched on a light. What it lit up was the pageantry - the decks, the pre-reads, the ceremonial alignment that let leaders look occupied while the actual decision sat in the corner checking its watch. None of that was leadership. It was leadership's stage set, and the set just became cheaper than the play.
It fooled everyone, including you, for a plain reason: people read effort as worth. Watch three days of visible toil and the output feels valuable, whatever it actually contained. Strip the toil away and the content has to stand there on its own, blinking, suddenly asked what it did all week.
So sort your week by one question: does this task carry a consequence only a person can own? Judgement under uncertainty, taste, the institutional memory of why the last deal soured, the signature at the bottom that means someone is accountable - keep those close. Summarising, formatting, comparing, moving information from one container to another - hand them over without ceremony. The machine can carry the procedure. It cannot carry the consequence.
Picture the partner who has to tell a ten-year client that the honest answer is no. The model can draft her the options, pull the account history, and stress-test every assumption before breakfast. It cannot decide which relationship can absorb the risk, which promise has to survive a bad quarter, or which easy revenue is not worth the scar. That call has a name on it, and the name is hers.
The work that survives is the work you would still have to answer for if it went wrong. Everything else was stationery.
Issue 00Two kinds of leaders are reading thisThe opener+
Two kinds of leadersare reading this.
One has let AI into the building and suspects they are using a spaceship to send email. The other has not started and is quietly hoping the whole thing blows over. The Signal is for both, and neither has to admit which they are.
There is a moment in almost every leadership team (or individual mind) right now where someone says "we / I need an AI strategy," everyone nods gravely, and not one person in the room could tell you what would actually be different on Monday if they had one. The word "strategy" is doing an enormous amount of unpaid labour. That gap, between sounding ready and being ready, is the only thing this newsletter is about.
Here is the uncomfortable part. Most of what you are sold as "AI transformation" is theatre: a percentage with no baseline, a partnership with more adjectives than workflow, a competitor's launch post engineered to ruin your Tuesday. The hype is loud, the evidence is shy, and the anxiety in between has quietly become someone else's marketing budget. Your nervous system is a line item in a funnel.
So let me help you calm your nervous system down. Do not start with a strategy offsite. Start with the one problem costing you the most sleep or time right now - the decision you keep circling, the bottleneck everyone tiptoes around, the one thing you keep doing yourself because "no one else can do it like me" or you "don't have the time to experiment or play" right now. Write your own view of it first, five plain bullets, before you open any AI. Literally, it takes a 10-minute coffee break you absolutely deserve. Then hand the model your thinking and tell it to attack: find the hole, the missing data, the strongest case against you. You are not asking the machine for the answer. You are using it to pressure-test the answer you already half-have. Do that once, on the thing that actually matters, and you will have done more real AI work than most companies manage in a quarter of pilots. Simple! Give it Context. Give it a role. Ask it to crit your work and interview you. Answer the questions and then give it the thing you want it to do based on that outcome.
That is the whole method here. Every Tuesday, one idea about AI and leadership that survives contact with a real week. No jargon, no maturity curves, no forecasting the technology as if it were weather. And every issue comes with a free companion - a one-page tool you can use on an actual decision, not a PDF you file and never open again.
If you have already started, The Signal will sharpen what you are doing and give you quiet permission to kill two or three projects you are keeping alive out of politeness. If you have not started, you are not behind - you are early enough to skip everyone else's expensive mistakes and bulldust baffles brains. Both are good seats to be in.
The only genuinely wrong move is to keep nodding in meetings, hoping nobody asks you what would be different on Monday, or to keep using the chat function only.
The Signal - AI and leadership, every Tuesday. Your data is yours; we never sell or share it.
