How I Improved My Storytelling GPT (And What’s Next)
About a year ago, I built a GPT to help professionals tell better stories. At the time, it was just okay—maybe a 3/10. The user experience wasn’t great, and the output felt mediocre at best.
Recently, I decided to go back and rework it, and now? I’d say it’s a solid 5 or 6/10. Still room for improvement, but it’s definitely better.
Along the way, I noticed a few surprising things about how it improved. I’ll share those, and at the end, I’ll talk about what it would take to push it to an 8 or 9/10.
Three Things That Surprised Me
1. Customized Examples (Without Being Told)
One of the biggest surprises was how the GPT started offering custom examples based on just a little input. I didn’t even explicitly program it to do this in the instructions or knowledge file—it just started behaving like a real coach. That was a huge leap from the earlier version.
2. Positive Reinforcement
I had asked the GPT to summarize the user’s input before giving feedback, but the way it did it felt natural—like a coach encouraging you. It didn’t just parrot back what was said; it framed it in a way that reinforced the storytelling process.
3. Recognizing Unanswered Questions
Another big improvement? It noticed when I skipped a question. For example, it asked, What was the end result of your efforts? and How did this experience change you personally?
I answered the first part but skipped the second, and instead of just repeating the question, it reworded it in a way that nudged me to answer: On a personal level, how did this change you?
The old version wouldn’t have done that.
How to Get from 6/10 to 9/10
Right now, I think this GPT has reached its limit. Adding more tweaks wouldn’t improve it much. Instead, the next step is to turn this into a structured workflow—a prompt chain.
A prompt chain is when you interact with an AI multiple times in a structured way to refine the output. This would mimic how a real storytelling coach works:
- Start with the basics – Get the bones of the story down.
- Refine key elements – Address common storytelling struggles.
- Add depth – Use AI to help with things like:
- Inner transformation – Many people struggle to highlight personal growth in their stories.
- Sensory details – People often tell instead of showing. AI can nudge them to add richer descriptions.
This kind of multi-step process would take the GPT from decent to actually helpful.
If you’re looking for a storytelling AI, I’ll drop the link to the GPT in the comments. Give it a try and let me know what you think!