Blog d'écriture IA : livres, paroles et création de contenu
How to Build a Manuscript Chain of Custody for Your Book
Build a practical manuscript chain of custody from origin through publication. VIP members can continue to the complete BUILD system.
Find Your Voice · APPLY
How to Build a Manuscript Chain of Custody for Your Book
Knowing that authors should preserve evidence is not enough. This guide shows you how to build a practical manuscript record for one real book—without turning your writing process into an administrative burden.
By the end, you will have a working folder that shows where the book began, how it developed, what AI contributed, what you changed, which sources you used, and who approved the finished manuscript.
Important: This is an educational documentation workflow, not legal advice. It does not guarantee copyright registration, publisher acceptance, platform approval, or a particular legal outcome.
What You Are Building
A manuscript chain of custody is the organized record showing how a book moved from the author’s original idea through research, outlining, drafting, AI assistance, rewriting, editing, source review, approval, and publication.
The goal is not to prove every keystroke. The goal is to make the development of the manuscript understandable.
Your finished basic record should answer seven questions
- Where did the project begin?
- Who directed the concept and structure?
- What tools and collaborators were involved?
- Which material was AI-generated, AI-assisted, or human-created?
- How did the manuscript change across major drafts?
- Which sources, permissions, and assets were used?
- Who approved the final version and what was published?
Step 1: Create One Master Project Folder
Do not scatter the record across random downloads, chat histories, email threads, and unnamed files. Create one master folder for the manuscript before you continue editing.
This folder can live in Google Drive, Dropbox, OneDrive, iCloud, a local drive with backup, or another system you trust. The storage tool matters less than consistency, access, and version control.
Do this now: Create the nine folders before moving to Step 2.
Step 2: Preserve the Project Origin
The strongest record begins before the polished manuscript exists. Save the material that shows where the project came from.
Save in 01_Project_Origin
- Original idea notes
- Handwritten pages or photographs of notebook pages
- Voice notes or transcripts
- Early working titles
- Reader problem or reader promise
- Personal experience behind the book
- Original framework, method, argument, story world, or teaching concept
- Date the project began
Do not rewrite these records to make them look cleaner. Their value is that they show the genuine starting point.
Simple origin note
Working title: The Book I Needed Five Years Ago
Project began: August 1, 2026
Reason for writing: I kept answering the same question from new creators and realized the answer needed a complete system.
Original human contribution: The central framework, the three-stage progression, the case examples, and the reader outcome.
Step 3: Record the Human-Created Structure
Structure is one of the clearest places where human direction becomes visible. Save the outline you intended to build, even if AI later helped expand or reorganize it.
Save in 02_Outline_and_Structure
- Reader profile
- Book promise
- Chapter list
- Section order
- Teaching progression or story arc
- Ideas that must remain
- Ideas intentionally excluded
- Reasoning behind major structural choices
When AI helps create an outline, preserve both the generated suggestion and the outline you approved. Label them differently.
Weak record
“AI helped me make the outline.”
Stronger record
“I supplied the audience, purpose, six required chapters, and final outcome. AI proposed three sequences. I selected one, moved Chapter 5 earlier, removed two generic sections, and added my original case study.”
Step 4: Build an AI Use Record
You do not need to archive every minor exchange. Record the uses that materially affected the book.
For each AI tool, record
- Tool name
- Date or project phase
- Purpose of use
- Material supplied by you
- Type of output produced
- Whether the output entered the manuscript
- What you changed or rejected
- Whether disclosure may be relevant
Example AI-use entry
Tool: ChatGPT
Task: Compare three possible chapter sequences.
Input: My human-created chapter list, reader promise, and required case studies.
Output used: One suggested sequence.
Human decisions: Reordered two chapters, rejected the proposed opening, added a new transition chapter, and rewrote all chapter descriptions.
Final status: Process assistance; no generated prose retained from this exchange.
Save important prompts when they establish concept, structure, research direction, chapter development, translation, image creation, or substantial revision. Do not confuse volume with quality. Five useful milestone records can explain more than thousands of unlabelled prompts.
Step 5: Save Milestone Drafts, Not Endless Copies
Version history should reveal development. It should not become a landfill of files named “final,” “final2,” and “final-really-final.”
Recommended milestone versions
- 01 Original Concept — premise, promise, framework, or story concept.
- 02 Human Outline — your intended structure before drafting.
- 03 AI-Assisted Development — material produced with tool support.
- 04 Working Draft — assembled manuscript before deep rewriting.
- 05 Human Rewrite — major voice, logic, story, and structure revision.
- 06 Editorial Review — editor or reviewer changes and comments.
- 07 Final Manuscript — approved publication text.
- 08 Published Edition — exact file or text released publicly.
Use dates and version numbers:
Step 6: Create a Decision Log
Drafts show that the text changed. A decision log explains why it changed. This is one of the most valuable records because it reveals authorship as judgment, not merely typing.
Record decisions such as
- Why a chapter was moved
- Why generated material was rejected
- Why a claim was corrected or removed
- Why a character, case study, or example was added
- How the voice was changed
- Why a source was accepted or replaced
- Why a section was condensed
- How the final expression differs from early output
Example decision entry
Date: September 12, 2026
Section: Chapter 4 opening
Problem: The draft began with a generic explanation and sounded detached from the reader’s experience.
Decision: Replaced it with my workshop story, moved the explanation after the story, and removed two unsupported claims.
Result: The chapter now begins with original experience and reaches the practical lesson faster.
Step 7: Capture Human Revision Evidence
You do not need a before-and-after record for every paragraph. Choose representative sections where your human contribution is substantial.
Use this four-part format
- Starting material: the AI-assisted or early draft.
- Human revision: the version you rewrote.
- Creative reason: why you changed it.
- Final contribution: what the revised section now accomplishes.
Good evidence may show changes in voice, argument, pacing, sequence, examples, factual accuracy, narrative meaning, emotional effect, or instructional clarity.
Do not manufacture evidence later. Save representative comparisons while the revision decisions are still fresh.
Step 8: Build the Source and Permission Record
A manuscript history is incomplete if it tracks AI but ignores third-party material.
Track in 06_Sources_and_Permissions
- Books, articles, papers, websites, interviews, and datasets
- Direct quotations
- Scripture editions or translations
- Public-domain texts and the edition used
- Images, charts, screenshots, and design assets
- Lyrics, poems, excerpts, and other copyrighted material
- Licences and written permissions
- Client, student, collaborator, or private material
- Facts requiring final verification
Use a simple spreadsheet with columns for source, creator, publication, link or identifier, material used, permission status, attribution requirement, and review notes.
Step 9: Record Editors, Collaborators, and Handoffs
A real book may pass through editors, proofreaders, translators, designers, researchers, sensitivity readers, ghostwriters, or co-authors. Record who handled the work and what they were responsible for.
For each collaborator, save
- Name and role
- Date the file was received
- Version supplied
- Work requested
- Changes returned
- What you accepted or rejected
- Agreement, licence, assignment, or permission where relevant
- Final approval responsibility
This is where manuscript chain of custody begins to connect with chain of title. The development record explains how the work changed. Agreements and permissions help establish who owns or controls the rights.
Step 10: Prepare the Final AI-Use Summary
Before publication, write a one-page private summary that answers the questions you may later be asked.
Final AI-use summary
- Which AI tools were used?
- What did each tool do?
- Was AI-generated text used in the final manuscript?
- Were AI-generated images or translations used?
- Which parts were independently created by the author?
- Which parts were substantially rewritten?
- Which outputs were rejected?
- Who reviewed and approved the final work?
- What platform or contractual disclosures were completed?
Keep this summary factual. Do not use vague statements such as “AI only helped a little.” Describe the actual tasks and outcomes.
Step 11: Freeze the Final Manuscript
When the manuscript is approved, create a locked final record.
Save in 08_Final_Manuscript
- Final editable manuscript
- Final PDF or submission file
- Final word count
- Approval date
- Name of approving author or publisher
- Final AI-use summary
- Final source and permission review
- Checksum, file hash, or read-only copy if your workflow supports it
The practical purpose is simple: months later, you should be able to identify the exact manuscript that was approved.
Step 12: Save the Publishing Record
The chain should continue through submission and publication.
Save in 09_Publishing_Record
- Submission dates and recipients
- Agent, editor, client, or publisher correspondence
- Platform upload notes
- AI disclosure answers
- ISBN, edition, and publication date
- Final cover and interior files
- Published URL or retailer listing
- Corrections made after publication
- New edition records
The 30-Minute Minimum Version
A creator with an existing manuscript may not be able to reconstruct everything. Do not invent records. Build the best honest record available.
- Create the nine-folder structure.
- Save the earliest genuine concept or draft you still have.
- Save three major manuscript versions.
- Write a one-page AI-use summary.
- List major sources and permissions.
- Save two representative before-and-after revisions.
- Record the final approval date and published file.
This is not as strong as documenting from the beginning, but it is better than relying entirely on memory.
Final Readiness Check
- The origin of the book is documented.
- The human-created structure is preserved.
- Major AI uses are recorded accurately.
- Milestone drafts are clearly named and dated.
- Important creative decisions are explained.
- Representative human revisions are preserved.
- Sources, assets, permissions, and licences are tracked.
- Editors and collaborators are documented.
- The final AI-use summary is complete.
- The approved manuscript is identifiable.
- Submission and publication records are saved.
- The folder is backed up.
What You Completed
You now have a working manuscript chain of custody.
You can explain where the book came from, how it developed, where AI entered the process, what you changed, what sources and collaborators were involved, and which file became the published work.
This public APPLY workflow creates the foundation. The gated BUILD stage turns it into a reusable professional system with structured logs, contribution evidence, source registers, approval sheets, disclosure preparation, and a final manuscript provenance file.
Related Reading
Final Word
A manuscript chain of custody is not about making writers defensive or turning creativity into surveillance. It is about preserving the story of the work while that story still exists.
Save the origin. Name the drafts. Record the decisions. Track the sources. Be honest about AI. Preserve the human work. Approve the final file.
That is how an author moves from saying, “I wrote this,” to being able to explain how the book became what it is.
Develop and publish the writing
AI can help produce words. Your voice decides what they mean.
Move from message and structure into a documented, complete and publishable project.
Discussion