These stories were written as a collaboration between Artificial Intelligence Large Language Models agents and me. I don’t want to claim I wrote them, because I very much didn’t. I suppose I commissioned them, and moulded them. I enjoyed reading them, so you might too.

My process is to have a long brainstorm with the agent about the subject matter. This helps me decide what I want, but also fills the agent’s context with source material. Sometimes I rig that context by copying material between agents.

Then I dictate a structure, in as much or as little detail as I like, and let it write a draft. Then I read it, ask for changes, make manual changes, and keep moulding until I’m happy with it.

I make some attempts to avoid a grating AI tone of voice, through a combination of hand-editing, providing examples, and prompting for tone. Otherwise, fooling the reader into believing AI had no hand in it, is not a goal.

In some stories I share some of my workings - partial transcripts or descriptions of the chats that produced the story. The wider terms for that practice are in The Kladde 26 Manifesto.

  1. Crab

    Sep 2026

    A papier-mâché crab becomes something much larger after the internet, and the state, misunderstand it.

    workings

  2. Calibration

    Sep 2026

    A magazine-style account of Perceptual Collapse Disorder — the Stare — and the printer calibration page that started it.

    workings

  3. A Man Who Believed in Fairies

    Jul 2026

    A man in a Northumberland pub tells a forensic analyst's story about AI chat logs, fairy folklore, and a missing schoolgirl.

    workings

  4. The Same Fire

    Jun 2026

    A story in two times — a rave in 1994 and a Calan Mai gathering in 1394.

    workings