This post was originally written in Italian by Paolo Gambardella. It has been translated by an AI agent and may contain inaccuracies.
Last week a colleague showed me an interesting video: a game dev getting ChatGPT’s Astra and Claude’s Fable 5.1 to work together inside Unreal Engine, to build a Souls-like boss fight.
The video itself is about game development, but what struck me was seeing two different AIs working on the same material rather than just one. It reminded me of a story I’d had in my head for a while — “Princess Melon”: I’ve been telling it to my daughter Maria Laura at bedtime since she was born, and over the past month I’ve actually sat down and written it up properly in my spare time. So I thought: why not try the same thing myself, with Claude and ChatGPT, but to finally put a face to its characters?
I spent two days getting the two services to talk to each other: one would propose a character, the other would refine or build on it, until we landed on a consistent cast across all the illustrations — from King Ant to the Little Mist, who feeds on the gossip of people who judge without looking. The process helped me understand who these characters really are, and gave me a much clearer picture of the story itself. When I showed Maria Laura the princess and the friends I tell her about every night, her jaw dropped: for the first time she could see them, not just hear about them.
This morning at the gym I was already thinking about the next step: getting in touch with a professional illustrator to give the book a proper, original look — and maybe, who knows, self-publishing my first fairy tale.
There’s a small irony in all of this. “Princess Melon” is about a Little Mist who thrives on “apparently…” and people who laugh before they’ve even bothered to look for themselves. When I read the comments from the usual modern-day luddites, quick to pass judgement on AI instead of actually trying it, I feel like I’ve somehow wandered into my own fairy tale.
