I’m about to publish my new book, The Bottleneck Detective. It’s a mix of fiction and non-fiction, written in a thoughtful, detective-noir voice. The book teaches people how to spot bottlenecks in the real world and turn businesses around. I write books for a living, but I had stopped because the physical strain of typing had become too much. This process changed everything. It improved the quality of my writing, made me more productive than ever, and helped me feel joyful about writing again.
My AI Workflow
I used three AI tools: ChatGPT, ClaudeAI, and Raycast AI Pro. These tools didn’t just speed up the process; they worked as creative partners. The experience reminded me of the Beatles writing music together—tossing ideas back and forth, each tool adding bits and pieces, and mixing everything up until I created something I loved.
But let me be clear: this wasn’t a case of me typing, “ChatGPT, write me a 10,000-word book on bottlenecks in the style of Encyclopedia Brown crossed with a textbook.” Who’d want to read that? The AI didn’t write my book for me. I spent dozens of hours working hard—thinking, shaping, rewriting, and collaborating with the tools to create something I’m proud of. The tools amplified my creativity, but the effort was all mine.
• I used ChatGPT for big-picture edits and untangling tricky sections. I pasted in whole chapters, reviewed its suggestions, and refined my ideas with its input.
• I leaned on ClaudeAI to help shape the tone. I started with a strong detective-noir voice but found it too “detective-y.” With Claude, I reshaped it into something more thoughtful and wise, with just a hint of noir. This process of refining the tone felt like uncovering the voice my book needed.
• I relied on Raycast AI Pro for extra capacity when the other tools hit their limits. Raycast works beautifully on Mac, but since I often write on my iPhone or iPad, I needed Claude and ChatGPT to fill the gaps.
The flexibility of these subscriptions helped a lot. When I needed more capacity, I subscribed to both Claude and ChatGPT. Now that I’m finishing up, I’ll cancel one of them and scale back.
Building the Book, Step by Step
This book began as a simple email course. Writing in small, manageable chunks helped me build momentum and gather feedback. The course focused on a single foundational story, with each lesson adding cumulative layers of insight.
When I finished the course, I decided to turn it into a book. I uploaded the text into ChatGPT and worked through it chapter by chapter. I refined, expanded, and sometimes completely rethought sections to make the book as strong as possible.
At one point, the AI suggested adding rules to complement the lessons. We came up with 25 rules. I reviewed each one, kept the best, and rewrote them to match the tone and focus of the book. This collaborative editing process helped me structure the book while keeping it clear and practical.
Why Grammarly (and Word) Didn’t Work
After finishing the manuscript, I ran it through Grammarly, hoping for help with final touches. Grammarly didn’t offer anything useful. It flagged issues that didn’t exist and even claimed a simple scene about drinking coffee was plagiarized. Its suggestions felt like noise, not signal.
Microsoft Word’s grammar tools struggled too. Both tools couldn’t handle the conversational, unconventional style I had worked hard to create. I canceled Grammarly because it added no value to this process.
The Human Touch
My wife read the book and spotted a subtle but important issue that none of the AI tools flagged. Fixing it took 2–3 hours but made a big difference. My proofreader is now polishing the text further, catching details no machine could.
AI and Creativity: A Joyful Collaboration
AI didn’t just speed up my process. It helped me think more clearly and creatively. I spoke drafts into ChatGPT and Claude, explored their suggestions, and refined the ideas into something better. These tools helped me untangle confusing sections, clarify my thoughts, and polish the manuscript.
This time, I spent less time hunched over a keyboard. Instead, I focused on shaping ideas and creating something I felt proud of. The reduction in physical strain felt liberating.
One of the most delightful parts of this process involved Eleven Labs. I used it to turn the manuscript into an audiobook-in-progress, which let me listen to the book while biking in the fresh air. I played it at 1.5x speed, and the voice sounded better than many professional audiobooks. Hearing the book in this way gave me a new perspective and helped me spot issues I might have missed on the page.
What I Learned
• AI tools can transform the writing process, but they work best as collaborators. The back-and-forth editing made all the difference.
• Tools like Claude and ChatGPT handled creative styles beautifully, but Grammarly and Word couldn’t keep up.
• Subscriptions let me scale up when I needed more capacity and scale back when I didn’t.
• Eleven Labs gave me a brilliant way to review my work by listening to it while enjoying the outdoors.
• Human input still plays an essential role in catching subtleties and making final improvements.
Writing this way felt incredible. The process didn’t just make me more productive—it let me focus on what I love most about writing: sharing helpful ideas. I can’t tell you how happy it makes me to know that the ideas in this book will reach readers instead of staying stuck in my head.
If you want to know more about these tools or how I used them, I’m happy to share.