History Co-Author Review

History Co-Author is an AI-assisted research and writing system created by Soren Jordansen, the developer behind Cozy Co-Author and Steamy Co-Author. Those earlier systems focused on fiction.

This is his first nonfiction system, built around a very different problem: how do you use AI to write history without ending up with generic prose, invented details, unsupported quotes or something that reads like a lightly rewritten Wikipedia article?

It runs inside Claude or ChatGPT and is built around three core files:

• Narrative History Framework
• Narrative History Style Pack
• The Setup Prompt

Together they guide the AI through research, planning, writing and book completion. There is also a Narrative History Studio for people using Claude Cowork or ChatGPT Work, plus a substantial training course covering the rest of the publishing process.

What I like about the concept is that it doesn't begin by asking AI to "write me a book about the War of 1812" and hoping for the best. There's a process.

Step 1: Plan

You add the system files to a Claude or ChatGPT project and run Setup. The AI then walks you through 18 planning questions, giving you options at each stage along with reasoning about what each choice will do to the finished book.

During Setup the AI pauses for three dedicated research stages. It goes to the sources, gathers material and builds a research record. The process eventually produces five planning documents, including a Research Bible containing the sources behind the book. That's a far more serious approach than asking a language model to rely on what it already "knows."

Step 2: Write

You then write the book chapter by chapter with the AI. The system has separate mechanisms for three common weaknesses in long AI manuscripts: voice, facts and continuity. The Framework controls the narrative approach, the Research Bible controls the factual foundation, and a running Book State tracks what has happened so later chapters don't lose the thread.

The result is supposed to feel like one coherent book rather than twenty separate AI responses stitched together.

Step 3: Publish

When the manuscript is complete you tell the system to "close out the book." It then creates supporting material:

• Cast of characters
• Note on names
• Note on sources
• Chapter notes
• Selected bibliography

The accompanying training then covers editing, formatting, covers, Kindle Direct Publishing, descriptions, keywords, promotion and series building. That makes it more than a writing framework. It's a path from "I have a historical subject I'm interested in" to "I have a published history book."

1. Narrative History Framework

The underlying architecture for the book. It includes what Soren calls an engine of consequence, along with controls for pace, scale and narrator style, plus rules for handling the historical record and quotations.

It was developed specifically for narrative history rather than generic nonfiction. That specialization matters: writing a practical gardening guide and writing an account of a military campaign involving documented people are fundamentally different jobs.

2. Narrative History Style Pack

AI writing has tells. Certain phrases turn up repeatedly. Paragraphs become too symmetrical. Every scene ends with a lesson. The prose becomes polished but strangely lifeless.

The Style Pack includes banned words and phrases, cliché controls, anti-AI writing rules and a self-diagnostic you can run on the output. It doesn't make every sentence brilliant. It means someone has deliberately addressed the habits that make AI nonfiction easy to recognize.

3. The Setup Prompt

Possibly the most important piece. Instead of handing you a giant instruction document, Setup conducts an interactive planning conversation. Those 18 steps define what kind of history you're writing, how it should be told, what the research needs to cover and which narrative choices guide the book.

This is why calling it a "prompt pack" isn't accurate. It's closer to an operating system for a history writing project.

Narrative History Studio

There's also a more automated option with two plugins, one for Claude Cowork and one for ChatGPT Work. They can create the book's folder structure, run Setup, start each chapter and run the final closeout. The Framework, Style Pack and Setup instructions are built in, so updates flow through without you replacing files manually. If you prefer working by hand, the original three files still work inside a normal project.

The nine narrator voices matter more than they sound

You can select from nine narrator voices or blend two. Narrative history isn't just facts: the same event can be told as a fast-moving military account, an intimate human story, a broad geopolitical narrative or a restrained documentary. You hear a sample before committing. That offers far more control than "write this in an engaging style."

You can also bring your own source material into the project, including letters, diaries and historical records, and have the system work from those first. For a serious history enthusiast, that could be the biggest attraction.

This is where any AI history product deserves extra scrutiny, because AI can produce extraordinarily convincing false information. History Co-Author doesn't claim to eliminate that risk entirely, and I think that's the right approach. Instead it builds safeguards into the writing process.

• Every quotation must come from a source, and is rechecked after the chapter is written.
• Figures are taken from the Research Bible.
• The system is instructed not to invent quotations, events, thoughts or dialogue for named historical figures.
• Where the record doesn't answer something, it writes around the uncertainty or tells the reader the record is unclear.

That's a much better standard than quietly filling gaps with plausible-sounding invention. Soren's own published example even includes source notes that explicitly acknowledge uncertainty. That is a positive sign, because real research often ends with "we don't know."

None of this means you should publish raw output. You're still the author. You still need to read your manuscript critically before putting your name on it.

Why this may beat generic AI prompting

Claude and ChatGPT are already very capable writers, so why buy a system that runs inside software you already have? Because the problem isn't getting AI to produce words. It's getting AI to research, structure and write an entire book consistently while following rules across hundreds of pages.

A cold prompt produces writing that sounds acceptable but has little texture. Events get summarized. People become names and dates. The AI hedges constantly. And unless required to work from sources, it writes from internal knowledge.

The sales page makes this concrete: the burning of Miami towns in 1790, written once with the system and once from a normal prompt. The system's version works from the general's report and includes documented details such as the number of houses and bushels of corn. The generic version never opened the report.

History Co-Author isn't mainly trying to make AI write more beautifully. It's trying to make AI research first, organize what it finds, preserve it and then write from it.

Soren used the system to research, write and publish a three-book series, Forgotten Wars of American Expansion, under the pen name Jonathan Hayes:

• The Army That Vanished
• Red Sticks Rising
• War in the Swamps

All three were researched, written and published in under fourteen days combined. Book 1 reached #1 in Amazon's History free chart, #1 in Native American History and #53 across the Kindle Store free chart during a free promotion. The "free" distinction matters. Those were not paid-store rankings.

The later results are arguably more interesting. Readers went on to buy Books 2 and 3. Red Sticks Rising received Amazon's #1 New Release badge in Native American History and reached #5 paid in History of Southern U.S. Books 2 and 3 also appeared together in the Native American History top 20. One reader found and bought Book 2 through search the day after upload, before it had been announced.

The two paid history books became the strongest earners in his KDP account, outperforming an existing cozy fantasy series. That isn't evidence every customer will reproduce those results. It is evidence the system has produced actual books that were published, discovered and bought by real readers.

There's also a useful quality test. Google's Gemini was given one of the books without being told how it was produced and asked whether AI wrote it. Gemini did identify AI involvement, but concluded:

"The average commercial reader will not know, nor will they suspect it."

That's more convincing to me than pretending a detector was completely fooled. The books are also available as paperbacks through Walmart.com.

An interesting detail about how it was developed

The research side apparently worked before the writing side did. Soren came close to shelving the project because the books were factually solid but dull. They read like Wikipedia, exactly the failure mode I'd expect. Instead of releasing it then, he kept working on the narrative framework until the material read like a story. That doesn't automatically make the prose exceptional, but it suggests the development focused on the right problem.

The training course

A 29-module course covers the whole publishing journey: choosing research material, planning the story, running Setup, narrative choices, writing, editing, formatting in Kindle Create, covers, KDP, descriptions, keywords, promotion and building a series. There are screenshot tutorials, video walkthroughs, action steps, progress tracking, discussion areas and FAQs.

For someone who has never published, that's valuable. For someone who already knows KDP, 29 modules may be more than necessary, though you can simply take the parts you need.

Why history is interesting for KDP

Narrative history is very different from the low-content, formula-driven categories that attract independent publishers. History books naturally form series, and each book can still tell a complete story. That suits Kindle Unlimited, where publishers earn from pages read. Soren's argument is that the narrative-history shelf is dominated by traditional publishers, academics and journalists, while few independents produce accessible, story-driven historical nonfiction for KU readers. That doesn't make history a guaranteed opportunity. It does make it an interesting one.

History Co-Author follows three earlier releases from Soren Jordansen, known as The Copy Viking. The milestones below are figures reported by the creator, not independently verified customer outcomes.

It starts with research instead of writing

Probably its biggest strength. Researching first and preserving the source base is fundamentally more sensible than generating prose and checking facts afterwards.

It acknowledges uncertainty

The framework doesn't instruct the AI to make every story neatly complete. Sometimes the record doesn't tell us what happened. Saying so beats inventing an answer.

It's built for long-form consistency

The Research Bible and Book State address two problems that get worse as manuscripts grow: forgotten facts and lost continuity.

It gives the user meaningful control

You still decide what kind of book you're making and how it sounds. The AI isn't supposed to disappear into a black box and return 60,000 words.

The creator has published books with it

Real books you can inspect on Amazon are far more useful than screenshots of hypothetical manuscripts in a chat window.

The publishing workflow is clear

Research, writing, finishing, formatting, KDP and promotion are treated as one process, which is helpful if you've bought writing software before but never reached the "Publish" button.

It's specifically for narrative history

The main system isn't trying to cover every nonfiction category. Biography, battle-focused history, ancient history and other specialized formats are separate frameworks. For some buyers that focus is a positive; for others it may feel restrictive.

You need a paid AI subscription

This isn't a standalone application. You need Claude or ChatGPT. If you already subscribe, it hardly matters. If not, it's an additional ongoing cost.

It isn't one-click publishing

You make decisions during planning. You read and edit the manuscript. You handle formatting, your cover, publishing and promotion. The training shows you how, but nothing uploads a finished book while you sleep. I prefer that, but buyers hunting for hands-off automation should know.

The course may feel overwhelming at first

Twenty-nine modules is substantial. A beginner will appreciate it. Someone who just wants to start writing may stare at the dashboard wondering where to begin.

Source quality still matters

The system can't create a rich historical record where none exists. If reliable sources are thin for your subject, it won't invent missing scenes. That's correct, but it makes certain subjects hard to turn into compelling books.

It's not a guarantee of publishing income

You still need to choose subjects people care about, produce good books, package them professionally and get them in front of readers. History Co-Author can provide a system. It can't guarantee an audience.

History enthusiasts who have always wanted to write a book

Possibly the most interesting audience. You may know more about the American Revolution, Ancient Rome, the Civil War, British royalty or World War II than anyone in your circle, yet organizing that into a publishable book feels impossible. A structured research and writing partner could make it realistic.

Aspiring publishers who keep buying tools but never publish

If your computer holds a collection of courses, prompts and unfinished KDP projects, a guided workflow beats another blank chat box.

Existing KDP publishers

You'll care less about the training and more about the framework. The real question is whether it produces materially better long-form nonfiction than your own prompting. Source management, the narrative framework and continuity tools make the strongest case.

Owners of Soren's previous Co-Author systems

The co-authoring philosophy will feel familiar. The major difference is the research discipline nonfiction history demands.

Who it's NOT for

• You expect instant passive income without publishing and marketing books.
• You don't want to edit AI-assisted writing before publication.
• You want completely automated, one-click book production.
• You want the AI to invent dialogue or fictional scenes involving real people.
• You specifically want to write historical fiction.
• You object to AI-assisted authorship on principle.
• You aren't willing to maintain a paid Claude or ChatGPT account.
• You want guaranteed factual perfection without checking your manuscript.

It's also worth asking whether you genuinely like history. The business model may be attractive, but researching and publishing history books becomes tedious fast if the subject itself doesn't interest you.

History Co-Author vs. just using ChatGPT or Claude yourself

Could you build something similar? Potentially, yes. A capable AI user could design research prompts, set citation rules, create style instructions, maintain a continuity file and build a multi-stage workflow. But that's not really the question. You could also build your own website and design your own cover.

The value depends on whether a completed framework saves enough experimentation and failed output to justify using someone else's system. The three files, guided Setup, Research Bible, Book State, narrator choices, source handling and publishing closeout are designed to work together. That integrated workflow is what you're buying.

History Co-Author is one of the more interesting AI publishing concepts I've seen, because it tackles a category where AI has both enormous potential and an obvious weakness.

The potential is speed: AI can read, organize and work with huge quantities of historical material far faster than a lone human researcher. The weakness is trust: history readers notice sloppy facts, and invented quotations can destroy a book's credibility.

The system seems to understand that tension. Instead of pretending AI never makes mistakes, it creates a disciplined research record, forces the writing to stay inside it and makes uncertainty visible when the evidence is incomplete. I also like that Soren has used it to publish an actual series rather than demonstrating it with sample chapters.

Does buying it automatically make you a successful history publisher? No. You'll still need judgment, editing, persistence and a willingness to finish and publish. But if you've wanted to create serious narrative history and the research and writing process has always stopped you, this is a thoughtful attempt to remove those barriers without pretending the author has no role left to play.

For existing KDP publishers, the attraction is a repeatable framework for entering a nonfiction category where research quality matters and generic prompting isn't enough. For history lovers who never thought of themselves as writers, the proposition is simpler still: the book you've thought about for years may no longer require you to become a professional historian before you can begin.

It includes a 14-day money-back guarantee, so if the concept fits what you want to do, you can look through the offer and decide whether the workflow makes sense for you.