Watch on Zoom instead
Watch on Zoom instead
Training resources
Stem Hooks skill file
- Download Stem Hooks V9, including full examples for Casual Yappers and Faceless AI Narrator Ads.
- Watch Stefan’s Stem Hooks guide to generate a batch, improve repetitive ideas, and save useful follow-ups as Snippets.
Reference videos
- Two Yapper examples — study the delivery, story structure, and conversational voice.
- Bonus: faceless AI narrator ad examples — additional references for another format; not covered on this call.
Follow Stefan’s chats
- V3 initial chat — follow the starting reference-ad workflow.
- V3 song ad — follow the song-ad adaptation.
- Norse Organics: turn a song ad into a Yapper — change the format while adapting the story to a different product.
- Adapt the Yapper to LURA and develop 20 more situations — vary the characters and situations while keeping the core emotion.
- 30 new LURA ad hook concepts — explore a wider set of openings.
- 30 Norse Organics hooks — compare how the hook exercise changes for another product.
Find a section
Apply the workflow to your product
- Start with relevant context and a strong reference. Give Chat the product facts it needs and an example of the format you want. Remove conflicting voice instructions that make the result sound unlike your reference.
- Approve an outline first. Check the situation, story, product connection, and close before asking for a full script.
- Refine the voice. Use examples of natural speech and give specific feedback on stiff or overly technical lines. Choose a creator whose delivery suits the story.
- Ask for 20 new situations. Keep the core emotion, but change the characters, setting, and events. Request a few sentences per concept so you can compare them quickly.
- Build a bank of 30 hooks. Push past generic openings. Pick ideas grounded in recognizable situations, then adapt a selected stem hook into the formats you want to test.
- Review before producing. Adapt references to your own product and audience. Check product claims and story details rather than treating generated copy as verified facts.

