- “GPT is incredible at creating logical structures, while Gemini excels at turning those structures into actual readable content.” — Jan and Olly
The problem
Olly's expert knowledge for his mentorship programme (Dream 100, welcome sequences, high-ticket coaching) was locked in his head, and turning it into comprehensive written training programs manually would be enormously slow. Generic AI-generated course content is useless without structure and real examples.
The AI solution
Olly sends 2-10 minute voice messages explaining his vision for a training. Jan transcribes them and feeds the transcript into GPT with massive, detailed prompts. GPT produces the structure: why the training matters, the mindset shifts needed, and step-by-step implementation. Olly then reviews and approves the outline before anything gets written.
The approved outline goes to a second model, Gemini, which writes the actual modules one by one. The split is deliberate: GPT is incredible at creating logical structures, while Gemini excels at turning those structures into readable content. Finally they add real case studies and examples from their own world — the human layer that makes the content valuable instead of generic.
The whole process is codified in a detailed SOP with well-thought-out prompts, so it runs repeatably rather than as one-off experiments.
Tools & costs
Results
Expert knowledge is turned into comprehensive training programs for Olly's mentorship programme from 2-10 minute voice memos; described as a game-changer for education businesses.
How to copy it
- Have the expert record a 2-10 minute voice message explaining the vision for one training.
- Transcribe it and feed it to GPT with a detailed prompt asking for structure: why it matters, mindset shifts, step-by-step implementation.
- Get the expert to review and approve the outline.
- Give the approved outline to Gemini and have it write the modules one by one.
- Add your own real case studies and examples, and save your prompts as an SOP.