Dan Martell was driving with his wife Renee when he pulled out the voice memo app, hit record, and spent 27 minutes answering one question out loud: in five years, what will be true? That conversation became raw material for an AI system that later separated the recording into two task lists and began executing. It sounds almost too simple, but that is precisely the point Martell has spent 15 years arriving at: the human moment comes first, and the machine processes what the human generates.
The stakes behind that drive are bigger than a to-do list. A marketing survey cited by Martell found that in 2025, 20 percent of consumers said heavy AI use made them trust a brand less. By 2026, that figure had doubled to 40 percent. Among Gen Z consumers who grew up with AI, the number jumped to 54 percent. People are not angry that AI exists; they are angry when they cannot find a real person behind the output.
Building the brain before writing a single word
The system Martell teaches has three layers, and the first one has nothing to do with prompts. It is about capturing everything: every meeting, every call, every decision, every AI chat. He uses a tool called Granola for meetings; others on his team carry a physical recording device or use a voice transcription tool to do end-of-day audio dumps. The tool, he is clear, does not matter. The capturing does.
All of it lands in one home base, whether that is Obsidian, Notion, or a Google Drive folder. From there, AI connectors pull in Gmail, Slack, Google Drive, and any content platform where the person has already published. Each connector takes roughly two minutes to set up. When Martell ran a session with 40 CEOs earning eight figures annually, a third of them had never connected their email to their AI system. That gap is the opportunity.
Once the connectors are live, a single interview prompt fills the remaining gaps. The AI is instructed to review everything it can access, then ask at least four rounds of questions about the person’s points of view, contrarian takes, and the stories behind their decisions. The output is a document called the brain system that travels into every future conversation.
The voice style guide that does the real work
The second document is where most people stop short. Martell fed the AI his 20 most recent sent emails, his 20 most recent Slack messages, and his 20 most recent documents, kept in separate buckets. The AI analyzed sentence length, punctuation habits, opening lines, closing lines, favorite words, and words never used. It found that his email voice, his Slack voice, and his scripting voice are genuinely different from each other, and it kept them that way.
He went further, screenshotting text messages and providing video outlines so the system could learn how he talks on camera versus how he types to a friend. As he put it: ‘I have never used an emoji in a script. You probably have never either, but you would in a text message reply.’ That level of granularity is what separates a voice style guide from a generic tone document.
Both documents, the brain system and the voice style guide, go into every chat going forward. But Martell is explicit that he reviews everything before it goes out. ‘I would never let something go out that was written by AI. That makes no sense to me.’ The system is for getting to a draft faster, not for removing the human from the output.
The MIT study he keeps returning to
Martell references an MIT study that hooked up 54 students writing essays and monitored brain activity. The AI-assisted group showed the weakest neural engagement of any group, and 83 percent of those students could not summarize their own essay minutes after finishing it. They had watched words appear without ever actually thinking them.
His framework addresses this directly: use AI to pull ideas from you, not push ideas onto you. Instead of asking the system what you should publish, tell it what is on your mind and ask it to help you align. The prompt he uses for a newsletter starts with: ‘Here is what I am thinking. Use my brain system and my voice style guide to make it sound exactly like me.’ The human supplies the ingredient; the system handles the packaging.
The 27-minute drive that started it all
Renee’s task list and Dan’s task list, sorted from a single car conversation, are still somewhere in the system Martell built. The brain parsed the two voices, separated the commitments, and sent them forward. That is the version of AI assistance he is describing: not a ghostwriter replacing a person, but a processing layer that takes what a person already thinks and moves it into the world faster.
For anyone who has spent 15 years creating content, the archive is already there. For anyone starting now, 30 days of consistent capturing, according to Martell, produces more raw material than most people generate in a year. The brain gets built either way. The question is whether the person feeding it stays in the driver’s seat.
