What People Write With AI: Use real creation patterns to choose a project with a clear job

Most people choose a format the way they choose a genre: by what sounds fun to write. The durable pattern is the opposite. Start by describing the reader’s job as an observable sequence of actions. Example: “The reader wants to learn a skill, practice it with short prompts, and track progress for 30 days.” That job doesn’t map to “a book” in general; it maps to an ebook plus exercises, and possibly a companion journal. When you define the job, the format becomes the container for the required actions—reading, doing, recording, reviewing—rather than a label. This is why formats repeat across successful AI-assisted projects: the reader behavior repeats.

If you are unsure whether a job needs a workbook or an ebook, use a simple test: if the reader must produce something (answers, checklists, meal plans, budgets, practice logs, reflection notes), the book needs a built-in production space. If the reader only needs explanations, a narrative ebook usually fits. If the reader needs both understanding and follow-through, the workbook/planner/journal family is the better match. You aren’t guessing based on trend; you’re matching a required behavior to a format that makes that behavior easy.

A format is not just “length and cover.” It is the amount of scaffolding the reader receives while using the content. Practical nonfiction often works as an ebook when the reader’s task is primarily comprehension and reference: instructions, explanations, examples, troubleshooting. Structured formats work when the reader’s task is primarily execution and iteration: repeated prompts, weekly planning, daily logs, recipe execution steps, or guided reflection. The same topic can demand different scaffolding depending on how the reader uses it.

Tradeoff to plan for: the more structure you add, the more review you owe. A cookbook that contains ingredient lists and step sequences must be internally consistent; a journal that asks readers to rate progress requires prompt clarity; a workbook that includes exercises requires questions that produce meaningful answers. If you choose a structured format for a topic that really only needs explanations, you increase page-work without increasing reader outcomes. If you choose an ebook for a topic that requires practice, you force the reader to do the missing work off-page and your book feels incomplete.

Narrow promise is the part that prevents format drift. A good narrow promise is specific enough that you can build a preview that proves the book works in its chosen container. Instead of “Strength training for beginners,” a narrower promise is closer to “A 4-week starter plan for runners who want to reduce injury risk and improve leg strength.” You should be able to show what the reader does in week one, what they record, and what changes by week four.

Your preview needs to validate three things: (1) the reader behavior you mapped to the format is actually supported (instructions include the steps the reader needs), (2) the promised scope fits the pages you plan (you can keep the plan coherent at the intended length), and (3) the voice stays consistent with the reader’s knowledge level. This is also where AI-assisted production becomes safer: you are not committing to the full pipeline until the format choice clearly serves the reader’s job.