Using ChatGPT as a personal editor

[jwcolby54]

My blog

I have been working on a book EventDrivenProgrammingInVBA which I am placing in a public GitHub so that anyone interested in the subject can get a free copy. Writing a book is a lot of work. Additionally getting editorial assistance is problematic. This is a free thing, so I cannot pay for such help, and folks who volunteer have their own lives to lead. Which means that getting the book edited is tough.

One day I asked ChatGPT to provide feedback on my Blog, and the results were interesting and contained many valid suggestions. Which caused me to think... how about ChatGPT as an editor. So I asked ChatGPT if it could do such a thing and well... it said yes. Understand that I have 190 pages written, I am not asking ChatGPT to write the book, but rather to perform editing duties.

EventDrivenProgrammingInVBA_ChatGPTEdited.pdf

It is interesting the way it works. First I tried to upload the entire book. The concept being that it could get more consistency across the whole. That didn't work! We agreed on what editing duties it would perform, and it launched into the process... and never came back. Which would become a pattern, but which got better as I got smarter about how to deal with ChatGPT.

First off, I split the process into parallel streams. In Firefox, I opened three tabs, and opened ChatGPT in each tab. I then uploaded a chapter into each tab. ChatGPT actually encouraged this paralleling of the process.

ChatGPT took direction as to what specifically to do. The following is what ChatGPT did for each chapter:

• Voice consistency:
  • Author = “I”
  • Reader = “you”
  • Adjusted consistently across explanations, instructions, and transitions
• Grammar and sentence structure:
  • Fixes for typos, awkward phrasing, run-ons, and unclear references
• Passive → active voice:
  • Converted only when it improves clarity or makes instructions more direct
  • Passive constructions left intact if stylistically intentional or more appropriate
• Flow and clarity:
  • Improved transitions, paragraph structure, and logical sequencing
  • Rewrites for readability and emphasis without altering meaning
• Preservation of formatting:
  • All fonts, sizes, headers, indentation, and code block formatting retained
  • Edits made inline to ensure seamless integration with your .fodt-based Git workflow
• Output format:
  • Final file saved and returned as .odt, ready for you to save-as .fodt

---

"I’ll apply this standard to all chapters unless you specify otherwise. Ready for the next chapter when you are."

I tried to have it do the edits in .fodt format. That ended up just not working reliably. .Fodt is xml. whereas .odt is a "zip" file with directories inside with stuff in the directories. For example graphic files are all stored in a directory inside the "zip" file' Other things as well. Text goes in a dir. Formatting in a different directory. My understanding of all this of course. I wanted to use .fodt because I wanted to to a/b comparisons between what I fed in and what I got back, using A program I found called Meld.

It just did not work. Too many niggling issues. After much struggle to stay with .fodt I switched to odt and it (mostly) just started working.

However it wasn't seamless. It would often hang. It would claim it was performing the edit but would never return. I would simply upload the same file in another tab and it would come back quickly. If it was going to perform the edits it would generally happen in 10 to 30 seconds. The larger the chapter, the longer the edit process.

It took two "days" to break my book down by chapters, feed each chapter into a tab (an upload process), wait for it to finish, download the edited file and save it in a staging directory. TBH I was not reading the returned chapters. In the middle I broke a couple of chapters in half.

Once all chapters were finished I reassembled the chapters into a new .dot document. I Built a new TOC.

There were very minor glitches. Several chapters where the first outline (chapter) level had a different font / pitch from what was fed in. A couple of places where the last paragraph was turned into a header level 3, the entire paragraph. These popped right out visually in the TOC where a paragraph was inserted into the TOC because it was level three formatting. I just had to go in and edit the paragraph to be body text and rebuild the TOC.

So... having a new document with all of that stuff "done" what is next?

1. Final Formatting Pass:
   • Scan for any:
     • Header inconsistencies (levels, spacing)
     • Orphaned lines or widows
     • Code formatting that got slightly shifted in merges
     • Page breaks between chapters (if needed)
2. Export Final PDF:
   • File → Export as PDF
   • Check:
     • TOC links are clickable
     • Margins and page size are as intended
     • Cover page (if any) is in place
   • Save a versioned copy like:
     EventDrivenProgrammingInVBA_<wbr>v1.0.pdf
3. Optional Extras:
   • Create a cover image (I can help design one)
   • Write a README or description for GitHub
   • Update your GitHub repository with:
     • Book.fodt
     • Book.pdf
     • All individual chapters (if desired)
     • License or usage notice

On top of that I still need the code itself pulled back out so I can insert it into a database to check for coding errors, as well as "sequencing" errors.

At any rate, my hope is that my time spent on this will make a real difference in readability.