Susan Frew, a Google AI Certified keynote speaker, took an unconventional path to publishing her business book "Recoded: Transform Your Business DNA with AI" by repurposing years of keynote recordings instead of starting with a blank page. She fed transcripts into NotebookLM and Claude AI, then built a complete multimedia product without hiring a ghostwriter or traditional publisher.

What Problem Does This Solve for Authors and Speakers?

Most speakers and business professionals already have valuable content locked in presentations and recordings. Frew's approach reveals a gap in the publishing workflow: speakers spend years perfecting material on stage, but that content rarely makes it into books. By treating keynote transcripts as a first draft rather than starting from scratch, authors can dramatically accelerate the writing process. This method also creates a natural quality filter, since stage-tested material has already been refined through audience feedback.

The workflow demonstrates how NotebookLM functions as more than a research tool. Frew used it as a "second brain" for the project, uploading transcripts and chapter drafts to ask questions about her own content, surface connections she missed, and identify gaps that needed more depth. Running NotebookLM alongside Claude AI gave her two complementary perspectives: one helped her write, while the other helped her think through the material.

How to Transform Your Keynotes Into a Published Book

• Record and Transcribe: Capture every live keynote and workshop session over time, then convert the audio to text using transcription tools. Gather all transcripts in one place before moving forward, without worrying about perfection at this stage.
• Organize With AI: Feed transcripts into Claude AI in manageable chunks and ask it to identify themes, key points, and natural chapter breaks. Work chapter by chapter rather than trying to process everything at once, then use Claude to sharpen language, add transitions, and tighten the narrative.
• Use NotebookLM for Synthesis: Upload all source transcripts and chapter drafts into NotebookLM to ask questions about your own content and explore it from different angles. Let it help identify gaps, themes, and areas needing more depth.
• Edit by Listening: Export chapter drafts into Descript and create a private podcast feed so you can listen to your chapters like episodes. Listen while walking or driving, and take notes on edits that jump out when you hear the words instead of reading them.
• Add Practical Value: Develop exercises and infographics for each chapter that help readers apply concepts to their own work. Keep exercises short, specific, and immediately actionable rather than just informational.

"Your ears are better editors than your eyes. Listening to your own book is one of the most underrated revision strategies in existence," noted Susan Frew.

Susan Frew, Google AI Certified Keynote Speaker and Author

This observation challenges the traditional approach of revising manuscripts on the page. When you listen to your writing, awkward phrasing jumps out, missing context becomes obvious, and sections that felt strong on paper suddenly sound thin when spoken aloud.

Where Does Voice Cloning Fit Into the Publishing Workflow?

The audiobook production step represents the most technically ambitious part of Frew's process. She cloned her own voice using 11 Labs, a voice synthesis platform, then brought the generated audio files into Descript for editing and syncing with the final published text. She taught herself how to remix the audio files and upload the finished audiobook directly to Audible through their ACX platform.

This approach eliminated the need for a recording studio, audio engineer, or professional voice actor. However, Frew acknowledged that voice cloning and audio remixing are not beginner-friendly tasks and may require hiring an audio editor if you're new to the process. The technical barrier is real, but for speakers comfortable with learning new tools, it opens the possibility of creating audiobooks without outsourcing the narration.

The complete toolkit Frew used included Claude AI for writing, NotebookLM for research and synthesis, Descript for audio editing and podcast creation, 11 Labs for voice cloning, Nano Banana for AI-generated cover art, Canva for design refinement, Fiverr for professional book formatting, Amazon KDP for print-on-demand publishing, and Audible's ACX platform for audiobook distribution. She hired a professional formatter on Fiverr for $700 total ($300 for print formatting and $400 for Kindle), which she identified as a worthwhile investment to avoid formatting errors that would require rework.

The broader implication of Frew's workflow is that the traditional publishing bottleneck has shifted. The hard part is no longer writing or even producing the book; it's marketing and distribution. By using AI tools strategically at each stage, speakers and subject matter experts can now move from raw content to finished multimedia products in a fraction of the time that traditional publishing requires, while maintaining full creative control and keeping all revenue from sales.