The challenge: a voice that did not scale
The client is a business consultant whose reputation rests on a distinct point of view. The problem was throughput. Writing to his own standard took so long that he published in bursts, then went quiet for weeks - and an audience that does not hear from you forgets you. He did not want generic AI writing; he wanted to sound like himself, more often.
What we built
We built a small set of AI tools around his actual material - his past articles, talks, and frameworks - rather than a generic writing assistant.
Tools trained on his own work
The tools draft in his structure and language because they are grounded in what he has already written and said. He edits rather than starts from a blank page, and the edits are about sharpening, not rescuing.
A path to the book
The same material and method fed a longer-form plan, turning scattered thinking into the spine of a book without losing the parts that make his work his.
The result
Output rose roughly fivefold while the voice stayed his. He now publishes on a steady cadence, with about half the time he used to spend, and the writing still reads as something only he would say.
Why it worked
The tools were built on his expertise, not a substitute for it. AI handled the mechanics of getting a first draft down; the judgement about what was worth saying stayed where it belongs.