The Missing Piece in AI Corporate Transformation Memos: CEOs

Every corporate AI memo asks employees to absorb a transformation. None of them include evidence the CEO has.

“AI is changing the way we work!” (still from “Last Exit to Springfield” - top five episode of the series)

Brian Armstrong sent Coinbase staff an email Tuesday telling them the company was cutting roughly 14% of headcount, partly because AI efficiencies had changed what the company needed to do. He is not the first CEO to make that call this year. He will not be the last.

There is a paragraph that shows up in nearly every one of these messages. Some version of: AI is changing the way we work. It is usually followed by a couple of examples: a team shipping faster, a function consolidating, a pilot that surprised everyone. The examples are real; they’re not inventing these things.

But almost every time, there is one piece of evidence missing. The CEO never tells you what AI has changed about their own job, their own habits and ways of working. That’s the reason this genre of CEO communications tends to hit as hollow. It presents as a classic “do as I say, not as I do” situation: these are great tools, for you.

The audience that matters most here is the remaining employees — the people who just watched colleagues clear their desks under the banner of AI efficiency, and are now being told to internalize the new pace. They are not stupid. They are doing exactly the math you would expect them to do: if this is so important that we just lost a seventh of the company, what is it changing about you?

When the answer is silence (or worse, abstraction) the message lands hollow. You are asking them to absorb a transformation you have not visibly absorbed yourself. That is the oldest authenticity gap in management. Putting AI on the marquee does not make it new.

That is where most CEO AI comms sits right now. The set is dressed, the cast is hitting marks, but you can tell the kitchen has never been used.¹ The audiences always can, though.

So I went looking for the opposite.

I spent parts of this week trying to find the substantive first-person accounts of tech CEOs² walking through how AI has actually changed their own week. The exercise was instructive, mostly because of what it returned.

The most-shared CEO AI memo of the last year, Tobi Lütke’s “reflexive AI usage” mandate at Shopify, tells the team that AI fluency is now a performance criterion and that no headcount request will be approved unless the requester proves the work cannot be done with a model first. It is muscular and unambiguous. It is also entirely about what other people must do. Lütke does not describe what he himself does, anywhere in the document.

The most-quoted CEO AI claim of last year — Marc Benioff announcing on the Logan Bartlett Show that AI now does “30% to 50%” of engineering, coding, and support work at Salesforce — is precisely the kind of macro number a layoff letter can be hung on. It contains no account of which 30% to 50% of Benioff’s time has changed.

The closest thing I found to an actual first-person workflow account was Jensen Huang last July on Fareed Zakaria GPS talking about querying multiple AIs in parallel and having them critique each other’s answers, like asking three doctors for opinions. He’s also said he uses chatbots for first drafts. That is real. It is also a sound bite, not really a structural shift.

The reported, workflow-specific account from a public-company CEO, the version that would let an employee actually feel like their leader is in this shift with them, has, as far as I can tell, not been written.³ But that also makes it a huge communications opportunity, internally and externally, for CEOs with concrete examples.

I had a version of this conversation with a client a couple months ago. The public-company CEO wanted more visibility in the AI-and-leadership discourse. The instinct was thought leadership: an op-ed, a podcast, a keynote, a thread on what AI means for the industry. My counsel: AI opinions are cheap. AI building and proof is novel and notable. Show how your own week has changed. Show the workflows you have broken and rebuilt. Show the things you can now do that you could not do six months ago, not as a flex, as evidence.

That is the part the Armstrong-style emails skip, and it is the part that would do the most work for them. A paragraph in those letters about “here is how my own job has changed, here is what I have stopped doing, here is what I now do that I could not,” would be worth more than every “AI is core to our future” sentence put together. It would also be the thing that gives the survivors permission to actually move, because somebody at the top did it first.

I’m not picking on Brian here, or not trying to. He just happens to be the CEO who sent this type of message this week. In fact, he is a real engineer-founder not that far removed from the non-CEO version of himself, the one who was writing code at Airbnb. The probability that he is not using AI heavily, right now, in interesting ways, is roughly zero.

Which is what makes the silence in the email so striking. This is a CEO who gets it, probably uses it extensively, and still wrote a 14% layoff letter that gives you no evidence of either. If you have ever managed people, you know what that reads as.

For CEOs about to make AI-driven workforce decisions (or more likely, the comms teams writing the letter on their behalf), there is no rhetorical workaround here. If you want the message to land as credible, you have to show you are doing as you say just as much as you are telling everyone else their lives are being transformed.

Otherwise the email is not a strategy. It is an anthropology paper about a company that happens to be yours.

1. Legendary 1970s Paramount head Robert Evans, in his memoirThe Kid Stays in the Picture, said on getting The Godfather right: the audience had to “smell the spaghetti” — you couldn’t fake an Italian-American family kitchen from the outside. I keep coming back to that phrase when I read these emails.

2. Not VCs. Sorry, Garry.

3. The case study in the opposite direction is Klarna, which began publicly replacing roughly 700 service workers with AI in early 2024, delivered Q1 2025 earnings via an AI-generated avatar of CEO Sebastian Siemiatkowski, and then quietly began rehiring humans through 2025 after admitting service quality had dropped.