CEO of an automation platform explains how he wrote a 168-page book with AI: 'It was like a Socratic dialogue'

Daniel Dines, the billionaire co-founder and CEO of UiPath, has turned his long-running habit of thinking with books into a book of his own. The result is The Work That Remains, a 168-page, roughly 40,000-word argument about how people and AI should split the labor: AI proposes, humans decide, automation carries it out.

Dines didn’t write it alone. He used Claude and ChatGPT, and he describes the process as closer to a Socratic dialogue than a one-way drafting tool. He would push ideas, test them against opposing views, and keep iterating. At times he even ran two models against each other with adversarial prompts, asking one to challenge the other’s theory and to react to articles he supplied.

The funny part is that the book got much longer before it got shorter. Dines said it swelled to nearly 100,000 words because the models were so verbose, and that the hardest part was cutting it back into something a reader could actually finish. He said the ideas in the book and his company strategy developed together over three years, while he kept working, reflecting, reading, and revising.

There’s a larger point behind the publishing stunt. Dines says AI is only as good as the data it was trained on, which makes it stronger on known problems than on genuinely new ideas. He also argues that harsh feedback matters for both humans and machines, because without it you drift into hallucination or routine. For him, the value of AI wasn’t that it wrote the book, but that it helped surface the book’s thinking.

Dines is refreshingly unromantic about the whole thing. He says he has ideas but not the time or the talent to put them into a form the public can use. That’s probably the most honest use case for AI on offer right now: not genius in a box, just a tireless first draft with decent manners.