Latest
Does Explicit Sex Sell Books? An Answer to The Absolute Degeneracy of Modern WritingThe Mary Sue: How a 1973 Star Trek Parody Became a Writing TermWhat the AI Writing Survey Says About Jobs, Freelancers and Fiction AuthorsHow Long Does Google Take? Crawling, Indexing and Recovery TimesLife of Pi50 Years of Duty: One Police Officer’s Compelling Journey from State Trooper to County SheriffBreaking Rank: A Top Cop’s Exposé of the Dark Side of American PolicingBlack in Blue: Lessons on Leadership, Breaking Barriers, and Racial ReconciliationCalled to Rise: A Life in Faithful Service to the Community That Made MeVigilance: My Life Serving America and Protecting Its Empire CityCollaborate or Perish! Reaching Across Boundaries in a Networked WorldTurnaround: How America’s Top Cop Reversed the Crime EpidemicThe Profession: A Memoir of Community, Race, and the Arc of Policing in AmericaSTEP-UP and LEAD: A Guide for Dynamic, Innovative Leadership in Law EnforcementStriving for Justice: A Black Sheriff in the Deep SouthThe Diagram GalleryGar Travis: A Tribute to One of the Best Faire PhotographersReverse Tropes: Six Ways to Turn a Tired Trope Inside OutA Hiker Used AI to Find a Waterfall. The Forum Sent Her to Hell.Did We Miscount the World’s Population? What the Rural Undercount MeansFrankenstein’s Creature Has Worn Many Faces. Only One Is Mary Shelley’sIs Consciousness Quantum? Penrose Is Half RightPetty Motivations Make Characters HumanAI Companies Are Preparing for a Catastrophe. Which One?Anatomy of a Writer: The Parts That Do the Real WorkGuillermo del Toro’s Writing Advice: Love Your MonstersFree Book Marketing: Seven Ways That Cost Time Instead of MoneyShould Your AI Report You to the Police?My Sister Built the Family We Never HadAnthropic Bans Cruelty Toward Claude: What It Means for WritersWork-for-Hire Contracts: What the Asimov’s Cover Fight Teaches FreelancersGenre Fiction vs Literary Fiction: Don’t Confuse Taste With SkillFlorida Hurricane Prep Rituals: The Grocery Run, the Water Pallet and the Generator in the BoxThe Most Insulting Line of Dialogue Ever Written for the ScreenLoki Through the Ages: From Norse Myth to Marvel, The Mask and Dogma“You Are Utterly Disgusting”: A Book Festival, an AI Cover Ban and a Pile-OnWho Rewrote the Sligachan Legend: AI or the Tour Buses?Layers: How I Ride Out Florida Power Outages in My ApartmentWhy I Don’t Like Reedsy for Ghostwriting: The NDA ProblemThe Enshittification of AmazonPublishers Cancel Books Over AI While Using It in SecretI Was Getting 100 Spam Emails a Day. $4.50 a Month Fixed It.World Mental Health Day: Nothing Was Wrong With MeKessler Syndrome: How Space Debris Could Close Earth’s OrbitAmazon Is Blocking Real Readers From Book ReviewsShould a Novella Get a Paperback, or Go Ebook Only?BookFunnel Download Problems: Fixes, Scams and AlternativesOzana Giusca: Reflecting the Architecture of IdentitySir Sean Connery: A TributeHow to Find Plot Holes in Your Novel (Most Are Character Holes)
The Writing King Your Ethical Ghostwriter. Your Story, Done Right.

AI for drafting the connective sections of your book

TL;DR: Chapter introductions, transitions between scenes and the setup paragraphs that move a reader from one idea to the next are the natural zone for AI drafting. They carry no specific voice and follow predictable patterns, so writing them by hand eats real hours. When the machine drafts them and a human rewrites them into voice, the workflow saves time without losing quality. More on the half-cost AI-assisted book and how AI fits into a book project.

I use AI every day as an assistant.

It cleans up transcripts, helps with session summaries, and roughs out outlines, and I still do the writing. Connective passages are the one kind of book prose I think a machine can rough out safely, because a chapter recap is plumbing and nobody buys a book for the plumbing. It’s misleading to hear “AI-assisted” used as though the machine wrote the chapters while the writer took the afternoon off.

What is connective work in a book?

Inside a finished book, certain stretches of prose aren’t distinctly the author’s: the two paragraphs at the start of chapter four that recap where the argument is and signal where it goes next, the bridge between an anecdote and the principle the anecdote illustrates, the setup paragraph that introduces a section and tells the reader why the next pages matter. Those passages keep the book flowing, but they’re not where a reader feels the author as a person on the page. They’re functional.

Any competent author of the genre could write them and they’d still read correctly. I don’t mind a machine drafting that kind of prose. The danger is how many writers can’t tell a transition from a story and hand the machine both.

Those connective stretches are perhaps fifteen to twenty-five percent of the words in a typical nonfiction book. When a writer drafts them by hand, they take real time, and the time produces no creative value, because the writer is just executing a structural function. That’s the work AI handles well. The recognition of which stretches qualify, and the drafting of them through the machine instead of by hand, is one of the larger time savings available on a book project.

Why are connective sections a good use of AI?

Three properties make connective work suited to AI drafting.

The first is that the prose is structurally predictable. A chapter intro recaps, sets up, and signals what comes next, in some order. Transitions acknowledge where the reader was, name where they’re going, and move them. Setup paragraphs orient the reader, raise the question, and point at the answer. These are forms, and the machine reproduces forms well.

The second is that the content of the connective stretches is mostly derivable from the surrounding material.

An intro to chapter four refers back to chapter three and forward to chapter five, both of which already exist in some form. The machine can read the surrounding chapters and produce a draft that fits structurally, because all the raw material is in the context. A third property is that the prose carries no voice. The reader doesn’t attribute these sentences to the author.

They’re reading them to get to the next stretch of voice-bearing material, and they don’t stop to enjoy the transition.

What is the two-step rule for AI drafting?

The discipline that makes this work is simple and easy to skip.

Step one is the machine drafts the connective section based on the surrounding voice-bearing material. Step two is a human reads the draft and rewrites it into the author’s voice, every time, with no exceptions. The rewrite is fast because the structural draft already exists. It’s necessary because the gear shift between a connective section in machine voice and a chapter in author voice is something a careful reader can feel even when they can’t name it.

Skip step two and the connective sections start sounding slightly different from the rest of the book. The voice of the author drops out for two paragraphs, then returns. Most readers can’t tell exactly what happened, but they can feel that something did. The book reads slightly choppier than it should. The stretches where the voice was absent dilute a reader’s investment in the author’s voice.

Every year I see more clients who tried to write their book with AI first and got a bad one. A draft like that doesn’t need to fall apart everywhere to lose the reader. An author who keeps leaving the room two paragraphs at a time is enough to make the reader stop trusting that anyone’s there.

The cumulative effect across a full manuscript is a book that feels less coherent than it should, and the culprit is the connective sections that nobody rewrote in voice.

What “rewrite into voice” means

The rewrite isn’t a thesaurus pass. It means reading the connective draft alongside the surrounding voice-bearing material and adjusting the prose until it fits. That means picking up the author’s typical sentence rhythms in this section of the book. The next step is replacing the machine’s generic phrasings with the specific phrases the author tends to use. Then catching any words the author wouldn’t use, even if they’re technically correct, and substituting the words the author would have used. Finally, tightening or loosening the sentence length to match the surrounding paragraphs.

This goes faster than original drafting because the structure’s already there.

A connective passage that would take an hour to draft from scratch takes ten or fifteen minutes to rewrite from an AI draft into voice. The savings come from skipping the structural work. Authors who try to skip the voice work as well end up with the gear-shift problem, and the hours they saved turn into a book readers put down halfway through.

Where this works well in practice

Chapter openings and closings benefit the most. The opening sets up what the chapter will cover, refers to the previous chapter, and motivates the read. The closing summarizes, gestures at the next chapter, and lands the section. Both are structural functions with predictable shapes, and both eat real time when drafted by hand on a book of two hundred pages.

Bridge paragraphs between major sections within a chapter are another good fit. The author has just told a story or made an argument, and the next section is going to take that material into a new direction. A bridge paragraph acknowledges what just happened, names what’s next, and moves the reader. For more on authentic voice over reinvention, see this profile of Kurt Russell. The machine draft handles the move competently, and the human rewrite makes it sound like the author.

Setup and recap paragraphs for technical material work well too.

A section that’s about to explain a concept needs an orientation paragraph that tells the reader what they’re about to learn and why it matters. A section that just covered a technical point needs a recap that consolidates the takeaways before moving on. Both are structural, both are predictable, and both run well through the machine with a human rewrite afterward.

When does using AI for connective sections fail?

The failure mode is gradual and easy to miss.

An author or writer starts using the machine for clear connective work, with disciplined rewrites in voice. The savings are real, the workflow holds, and the book progresses. Then a connective section feels like it could just be ten percent voice-bearing, and the writer lets the machine draft it anyway, with a lighter rewrite. Then a longer setup paragraph that’s closer to argument than to scaffolding gets the AI treatment, again with a lighter touch.

Within a few chapters, the line has drifted, and the machine is drafting voice-bearing material that a human edits instead of writes.

I don’t have patience for this drift, because the writer knows where the line is and walks over it anyway. It’s a deadline decision wearing a craft costume, and the client’s the one who pays for it with a book that sounds like nobody in particular.

The drift produces the flat result that AI-assisted projects are notorious for when they fail. The fix is the discipline of the original two-step rule. Connective sections, as defined narrowly above, get the AI draft. Voice-bearing material, including anything close to argument or anecdote or specific perspective, gets a human draft from scratch. The line stays where it should be, the savings come from the connective work, and the quality comes from the voice work.

The anchor piece on the voice rule covers the broader version of this discipline, and the discipline is the same on every section of the book.

Hand the machine the plumbing and keep a human on everything a reader will remember. A writer who lets AI draft the stories, arguments and opinions in your book and calls the light pass afterward “editing” is selling you a cheaper product at full price, and you have every right to ask about it before you sign.

Frequently Asked Questions

What counts as connective writing in a nonfiction book?
I define connective sections as the parts of a book that carry the structure instead of my own voice. These include chapter openings and closings, bridges between major sections within a chapter, and setup or recap paragraphs for technical material. They recap where the argument stood, signal where it’s going next, or orient the reader before the next point. In a typical nonfiction book these stretches make up roughly fifteen to twenty five percent of the words, and they follow predictable patterns that any competent author in the genre could write. I let AI draft this material because it carries no specific voice, then I rewrite it into my own voice before it goes into the manuscript.
How much of a book is connective?
Roughly fifteen to twenty-five percent of the words in a typical nonfiction book. Enough that automating the drafting saves real time. Small enough that the voice-bearing eighty-plus percent stays in human hands where it belongs.
What’s the two-step rule for drafting connective sections with AI?
Step one is the machine drafts the connective section based on the surrounding voice-bearing material. Step two is a human reads the draft and rewrites it into the author’s voice, every time, with no exceptions. Skip step two and the gear-shift problem ruins AI-assisted books.
Why can’t I just skip the rewrite?
Because readers can feel the gear shift between a connective section in machine voice and a chapter in author voice, even when they can’t name what they noticed. The cumulative effect across a manuscript is a book that feels less coherent. The rewrite is what prevents that.
How do I know if a section is connective or voice-bearing?
Ask whether a reader will attribute the prose to the author. If yes, it’s voice-bearing and stays in human hands from the first draft. If the prose is structurally functional and any competent writer could have produced it correctly, it’s connective and the machine can draft it for a human to rewrite into voice.

The book on this: The Ghostwriting Advantage is 254 pages on which parts of a book need a human and which never did. That’s the distinction the connective sections sit on.

About the Author
Richard Lowe, professional ghostwriter

Richard Lowe is a professional ghostwriter and author with 113+ books authored and 54+ ghostwritten. Before writing full time he spent 33 years in enterprise technology, including 20 years as Director of Computer Operations and Technical Services at Trader Joe's. He writes nonfiction, fiction and memoir, and works with executives and experts on books that build authority.

More about Richard Lowe →

Disclaimer

The views and opinions expressed in this blog post are solely those of Richard Lowe and are based on personal experience and research. This content is for informational purposes only and should not be construed as professional legal, financial, accounting, or business advice. Always consult with qualified professionals before making important business or legal decisions. Richard Lowe is not a lawyer, accountant, or licensed professional advisor, and this content does not establish any professional relationship.

0 comments

No comments yet. Yours can be the first.

Was this useful?

Leave a comment