Which model do you use for working on documents, and how do you handle versions and revisions?

Hi everyone,
I work regularly on business documentation, information system docs, and other documents, and I’m curious about how people here use ChatGPT in these situations. I have Plus plan.

Pain points:

• When working on a document and asking the model for a small change (like updating a sentence), sometimes it rewrites or changes other parts of the document I was already happy with.
• Because of this, I often save interim versions as backups in case too much content gets changed, so I can roll back if needed.
• I’m wondering if this is just the nature of the model, or if there are ways to get more precise changes.
• I understand that some models rely more on semantic context (chunking the data), while others might be better at remembering literal text.

Things I’m interested in:

1. Which model or custom instructions do you use when working on documents where you want minimal unintended changes?
2. What workflow do you follow to manage document versions and revisions when using ChatGPT?
3. Have you found any prompt techniques or templates that help keep changes targeted and avoid rewriting more than you asked for?
4. Do you ever use “delta” instructions, like “Only change this one part, don’t touch anything else”, and does that work for you?

A small change shouldn’t introduce a lot of unintended rewriting or drift in terminology. If the model rewrites large portions, it forces extra work and can undermine any version control discipline.

Would love to hear what works for you!