AIWriteBook Authors in 2026: The Data Behind 13,683 Books

01

Who actually writes with AIWriteBook

Of 10,281 registered users, 9,266 — 90% — created at least one book. That is an unusually thin layer of window shoppers: people who sign up almost always start a manuscript. What happens after that first book is where the population splits.

Two very different authors live in this table. The 81% majority is writing THE book — the memoir, the novel, the guide they have carried around for years. And at the far end sit 62 serial publishers with eleven or more books each, treating the platform as a production line. Both are real, and the product decisions that serve one often bore the other.

Honesty requires the funnel too. Among the 6,767 engaged writers from January through early June 2026, 62% generated at least one chapter, 12% completed an export, 1.4% generated KDP metadata, and 0.6% cross-published to our reading platform. The pipeline narrows hard after the writing itself — finishing and shipping a book remains the exception, not the rule, AI or no AI.

02

The 2.2-hour draft

Among books that were completed, the median time from creating the book to generating its final chapter was roughly 2.2 hours. Read that carefully: half of all finished manuscripts on the platform went from empty project to full draft in a single evening.

For context, a first draft has traditionally been a months-long undertaking — NaNoWriMo built an entire global movement around the challenge of drafting 50,000 words in thirty days, and plenty of working novelists consider a year per draft normal. A 2.2-hour median does not make those authors wrong; it means the drafting stage specifically has collapsed, while everything around it has not.

Two caveats keep this honest. First, 2.2 hours is drafting, not publishing — editing, proofreading, cover design, and the KDP listing still take real hours or days. Second, the median is measured among finishers, the people who pushed through every chapter; it says nothing about the majority who stall mid-outline. The mechanism behind the speed is the outline-first loop: you approve a chapter-by-chapter plan, then the chapter generator writes against it one chapter at a time, so a committed evening genuinely can produce a complete draft.

03

Beyond English: 39 writing languages

English dominates, but less than you might guess: 11,056 of 13,683 books, or about 81%. Roughly one in five books on the platform is written in another language — 39 languages in total appeared in 2026.

The Spanish figure combines Spain (439 books) and Mexico (181). The long tail past this table runs through thirty more languages, from Norwegian to Hindi — mostly authors writing for their home market rather than translating into it.

That is the pattern AIWriteBook is built for: the whole flow — outline, characters, editor, export — runs natively in 30+ languages, so a writer in São Paulo drafts in Portuguese from the first word instead of writing English and converting later. And once a book is finished, the whole-book translation tool turns one manuscript into editions for other markets in a single pass, which is how several of the multilingual authors in this dataset ended up with the same title in three languages.

04

From draft to shelf

Export is where a project stops being a document and starts being a book, and the export numbers show heavy iteration: 5,508 fiction export completions by roughly 741 distinct writers, and 1,837 non-fiction exports by roughly 350. That is seven-plus exports per exporting author — people generate, read the file, revise, and export again, exactly the way you would proof a paperback. The most common formats are print-ready PDF, EPUB, and DOC: the trio an Amazon KDP upload actually needs.

Public publishing is smaller and worth stating precisely. 191 books have been published openly on NanoReads, AIWriteBook's reading platform, and 6,869 author pages are active. 470 books are publicly showcased in the books written with AIWriteBook gallery — 347 fiction, 123 non-fiction — and 38 of those carry live Amazon buy links.

Those 38 are the books whose authors chose to attach a store link; more have certainly been listed quietly, and we have no visibility into what any of them sell. If you are weighing what a listed book can actually earn per copy, the arithmetic lives in our KDP royalty breakdown, and it is sobering enough that you should read it before you price anything.

The listing itself is the step our data says writers skip most — only 1.4% of engaged writers generated KDP metadata, even though keywords and categories decide whether a book is ever found. If that is the step you are avoiding, the free KDP keyword research tool does the uncomfortable part in a few minutes.

One emerging niche deserves a mention with its true size attached: audio. In 2026, 939 chapters were narrated and 74 audiobook exports were completed — by just 30 users. Small in headcount, deep in usage: the writers who convert a finished book with the audiobook generator tend to convert several. It is a pattern worth watching, not yet a movement.

05

The import crowd: 3,291 manuscripts brought in

Not everyone starts from a blank page. Writers imported 3,291 existing manuscripts in the study window — drafts written in Word, Google Docs, or another tool, often years old, brought in to be finished rather than started.

This group behaves differently from first-timers. They arrive with the hardest part — a partial book they believe in — already done, and they use the platform to break the specific wall that stalled them: a saggy middle, a rewrite they kept postponing, a finished draft that needed conversion into an audiobook or another language. Draft import preserves the original chapters and extracts characters and structure from them, so the AI continues the author's book instead of replacing it.

If there is one segment of this dataset we would point a stalled writer toward, it is this one. A manuscript that has survived years in a drawer has already proven the author cares about it; the import path exists so that caring is the only prerequisite left.

Questions this data can answer