kryme.ai

kryme.ai

Kryme Monthly Letter — August 2026

The Outline Problem: When AI Helps—and When It Hurts

kryme.ai's avatar
kryme.ai
Aug 30, 2026
∙ Paid

Letter from the Editor

To the Writers of Kryme.ai,

There is a question I have been thinking about quite a bit lately, partly because of where artificial intelligence is heading and partly because of the way crime writers are beginning to use it.

Should you use an AI prompt to outline a novel?

A year or two ago, I might have answered that question very differently. The first generation of writing prompts was mostly about convenience. You would tell an AI that you were writing a murder mystery, give it a few details about the protagonist and setting, ask for thirty chapters, and wait while it produced something that looked remarkably like a complete outline. There would be a body in the opening chapters, several suspects, a few secrets, an apparent breakthrough somewhere near the midpoint, a setback late in the book, and a confrontation with the killer at the end.

If you had never outlined a crime novel before, the result could be impressive simply because there was so much of it. Thirty numbered chapters have a certain authority when they appear on the screen all at once. The story looks substantial. It looks organized. Most importantly, it looks finished enough that you can imagine sitting down the next morning and beginning Chapter One.

Unfortunately, appearance is one of the great dangers in both crime fiction and artificial intelligence.

An outline can contain thirty chapters and still have no real investigation inside it. It can contain six suspects without giving any of them a credible reason to remain under suspicion. It can contain fifteen clues without establishing where those clues came from, why they were not discovered earlier, or what they actually prove. It can contain a spectacular final twist that depends on information the reader never had a chance to know. It can move the protagonist from discovery to interview to surveillance footage to forensic report so efficiently that nobody stops to ask whether any of those discoveries logically caused the next one.

This is where the question becomes more interesting, because I do not believe the conclusion should be that AI prompts are bad for outlining.

I think the conclusion is that most of us have been asking too little of them.

Keep reading with a 7-day free trial

Subscribe to kryme.ai to keep reading this post and get 7 days of free access to the full post archives.

Already a paid subscriber? Sign in
© 2026 Mike Dickson · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture