I Spent a Day Writing a Short Story with Claude.
I had an idea for a science fiction story. A device that gives you the best version of yourself for 72 hours, then kills you. Only eight exist. What would people do with them?
I wanted to explore the allegory that life itself is that device. We're all given extraordinary capabilities for a limited time, and most of us waste them on the same things the characters in the story waste theirs on: status, money, vindication, pleasure. The question was whether I could turn that into a story that actually worked as fiction, not just as a thought experiment.
I can't write fiction. I've never written a short story. So I collaborated with Claude. I came up with the premise, the characters, the plot structure, and made every creative decision. Claude generated the prose. We went through six rounds of editorial review using AI subagents acting as a sci-fi editor, a prose editor, a story structure critic, and a general sci-fi reader. I caught inconsistencies, pushed back on suggestions, changed direction multiple times, and shaped every element of the world and the characters.
The result is a 10,000-word short story called "Seventy-Two." You can read it here: https://marcusvero.com
I don't know if it's good. That's partly why I'm publishing it: to find out. If you read fiction, I'd genuinely like to know what you think. Did the ending land? Did you finish it? Where did you stop caring?
If you're more interested in the process than the story, I'll write about that separately: how the collaboration worked, what the AI was good at, what it was bad at, and what surprised me. But the story should stand on its own first.
I had an idea for a science fiction story. A device that gives you the best version of yourself for 72 hours, then kills you. Only eight exist. What would people do with them?
I wanted to explore the allegory that life itself is that device. We're all given extraordinary capabilities for a limited time, and most of us waste them on the same things the characters in the story waste theirs on: status, money, vindication, pleasure. The question was whether I could turn that into a story that actually worked as fiction, not just as a thought experiment.
I can't write fiction. I've never written a short story. So I collaborated with Claude. I came up with the premise, the characters, the plot structure, and made every creative decision. Claude generated the prose. We went through six rounds of editorial review using AI subagents acting as a sci-fi editor, a prose editor, a story structure critic, and a general sci-fi reader. I caught inconsistencies, pushed back on suggestions, changed direction multiple times, and shaped every element of the world and the characters.
The result is a 10,000-word short story called "Seventy-Two." You can read it here: https://marcusvero.com
I don't know if it's good. That's partly why I'm publishing it: to find out. If you read fiction, I'd genuinely like to know what you think. Did the ending land? Did you finish it? Where did you stop caring?
If you're more interested in the process than the story, I'll write about that separately: how the collaboration worked, what the AI was good at, what it was bad at, and what surprised me. But the story should stand on its own first.
111❤️9💜2🤙2🤔1
