#4 | He said it out loud, but never wrote it down

On March 26, 2026, Notion launched a campaign called "Think Together."

I have been following Ivan Zhao, Notion's co-founder, since around 2022. He is the one who spent a month in Kyoto rewriting Notion from the inside, and he has talked about that experience in a lot of places since.

"Think Together" landed the way I expected it to land. It sits in a lineage of three-word rallying cries: IBM's "Think" in 1930, Apple's "Think Different" in 1997, and now Notion's "Think Together" in 2026. That framing is very Ivan. So that day, I went looking for the most recent long-form piece Notion had published under his name, and I opened it braced for a manifesto. The title was "Steam, Steel, and Infinite Minds," published in December 2025. A long essay about AI and the future of Notion, with Ivan's name on it.

I was expecting certain words to show up.

Five of them, actually. craft. editor. editorial. taste. curation. These are the words Ivan has reached for, over the years, whenever he talks about what Notion is trying to be. So an AI essay under his name would probably carry at least a few of them. That was my assumption.

I searched the body text.

Not one of them. Five for five.

What was in the body instead was a different register entirely: steel, steam, infinite minds, agents, Carnegie, Red Flag Act, a line about 30–40× engineer. An essay about AI, written in the vocabulary of the nineteenth-century industrial revolution.

The one word that did show up was human scale, twice. Both instances framed it as something that will not survive — "human scale does not hold in an AI era." So even the closest word was pointed in the opposite direction from what I had expected.

I'll be honest, I doubted my own search first. Maybe he was using synonyms. I went back and looked again. No. The essay Ivan put his name on, about AI, contains zero of the five words that used to define how he talked about Notion.

So I stopped for a moment.

The most obvious reading is: Ivan changed his mind at the end of 2025. Faced with AI, he pivoted from thinker to industrialist. Read that way, the essay makes sense on its own terms.

But that reading has to explain one other thing that happened during the same four months.

In late February 2026, Ivan went on "Possible," the podcast run by Reid Hoffman. The official Apple Podcasts episode description for that episode is careful about what it says Ivan actually talked about. Two phrases it pulls forward:

"in an AI-powered future, human judgment, taste, and values matter most"

"human scale in an AI-driven world"

Every word that was missing from the essay was there on the podcast. Not as filler. The description foregrounds human judgment, taste, values, human scale — the exact set of words that would have defined the essay, if the essay had been written the way I expected.

I have not heard the full audio, because the episode's primary feed sits behind a paywall I have not crossed. So I cannot tell you where inside the conversation Ivan said those words, or in what context. What I can tell you is that the show's producers — the people whose job it is to summarize what Ivan actually said in that episode — reached for exactly those four words when they had to write the official description. That is not a mistranslation or a loose paraphrase. That is the show pulling out what the show decided the episode was about.

The same person, over the same four months, used words in speech that he never used in writing.

And the binary most people reach for here, "did Ivan change his mind or not," doesn't quite catch what is going on.

Neither does the other common framing, "is Think Together an editorial manifesto or an AI strategy announcement?" Read only the essay and it is an AI strategy announcement. Read only the podcast and it is an editorial manifesto. Both readings are half right, and each one only survives by ignoring the other.

The real question is different.

Why did he use one set of words when he wrote, and a different set of words when he spoke?

And what does that asymmetry, by itself, tell us about how to read anything a CEO says from now on?

The rest of this issue is behind the paywall. It is not just a continuation of the observation. I want to turn it into a method — a way of reading CEO communication that you can apply to Altman, Amodei, Nadella, or anyone whose posture you are trying to figure out before you commit to their product. A three-step reading technique, and then a re-reading of the whole Think Together campaign as a single campaign edited across four media layers instead of one.