I had a great interaction with a user today about why they blew through their limits in just a handful of conversations whereas the previous period they had plenty of time and interaction. Turns out they were locked into the Storyforming Agent even beyond the point where they had figure the Storyform out!
Anyone who has been a subscriber of Subtxt knows the long journey it has been to get everything from GPT-3.5 to the latest GPT-5 to adhere to Dramatica theory. With 5.2 and our new Storyform Builder I can confidently say that this “problem” has been solved.
It just takes a lot of compute!
The conversation inspired me to write up an article about the difference between the Storyforming Agent and all the others, and so I hope this gives you a better idea of how everything works:
Why Storyforming Is Computationally Heavy (and Why That’s a Feature, Not a Bug)