TL;DR - If ChatGPT already gave you a strong draft, the problem is usually
not writing anymore. The problem is handoff. Ask ChatGPT for Markdown only, do
a quick cleanup pass, then use depapel to turn that
Markdown into a shareable PDF or
DOCX. Keep ChatGPT for drafting. Use depapel for the file
you actually send.
ChatGPT is often the place where the first usable draft appears.
That is useful while you are still drafting and refining the content.
The friction starts later. A client wants the brief as a PDF. A manager wants a
Word doc with comments. A teammate wants a file attachment instead of another
chat link. At that point, the task is no longer "use ChatGPT well." The task is
"turn this Markdown draft into a document other people can use."
The practical workflow
Use this order:
- Ask ChatGPT for Markdown only.
- Remove wrapper text and fix obvious structure issues.
- Paste the cleaned Markdown into depapel.
- Export PDF when the file should look fixed.
- Export DOCX when the file still needs comments or edits.
That is the whole workflow. It is simple on purpose.
Why the default copy-paste path breaks down
The weak version of this workflow looks like this:
- Ask ChatGPT for a draft.
- Copy the whole answer into email, Google Docs, or a chat message.
- Fix formatting manually.
- Hope the result still looks clean when shared.
That usually creates unnecessary cleanup work.
Common problems:
- ChatGPT adds wrapper lines like "Here is the Markdown".
- Heading levels skip from
# to ###.
- The whole document arrives inside one outer code fence.
- A wide table should really have been a bullet list.
- Code blocks are missing language tags.
- The final file is shared as a screenshot, not a reusable document.
The fix is not a complicated prompt library. The fix is to keep the Markdown
clean before it becomes a file.
What to ask ChatGPT for
The prompt does not need to be fancy. It just needs a strong shape.
Write the output in Markdown only.
Do not add any intro sentence or outer code fence.
Use one H1 title.
Use ## for main sections.
Use bullet lists for steps and options.
Use fenced code blocks with language tags when needed.
Prefer short paragraphs.
Avoid wide tables unless the comparison really needs columns.
This will be exported to PDF or DOCX, so keep the structure clean.
That removes most of the formatting noise before it starts.
A concrete example
This is the kind of raw Markdown output from ChatGPT that is easy to convert
cleanly:
# Launch Brief
## Summary
Ship the updated onboarding guide this week.
## Risks
- Reviewer comments may arrive after design freeze.
- Large tables should be checked before final PDF export.
## Delivery options
| Format | Use when |
|---|---|
| PDF | The document should look final |
| DOCX | The reviewer wants comments and edits |
## Next step
```text
Export target: PDF for delivery, DOCX for review.
```
This works because the source is already document-shaped. It has a title,
headings, a short list, a small table, and a tagged code block. That gives the
export step something stable to work with.
Where depapel fits
Once the Markdown is clean, depapel is the export layer:
- paste the ChatGPT Markdown into the editor,
- upload a
.md file if you already saved it locally.
depapel is a fit here because the same Markdown can become either:
- PDF for fixed delivery, or
- DOCX for review and edits.
That is the useful boundary. depapel does not connect to your ChatGPT account,
fetch private chats, or run a model-specific integration. It converts the
Markdown you provide.
When to use something else
Keep working inside ChatGPT when:
- the draft is still changing heavily,
- the output is still exploratory,
- or you do not need a file yet.
Use depapel when:
- the content is stable enough to share,
- someone needs an attachment instead of a chat,
- you want PDF and DOCX from the same source,
- or you need a cleaner handoff than copy-paste formatting can give you.
Honest limitations
This workflow still has limits.
- depapel does not pull content directly from ChatGPT.
- You still need to review AI-generated structure before exporting.
- Complex embedded HTML, advanced equations, or messy citations may need manual
cleanup first.
- If the source Markdown is poor, the export step will not magically make it a
well-edited document.
That is why the cleanup pass matters.
Final rule to remember
Use ChatGPT to draft the Markdown.
Use depapel to package that Markdown into the file you actually need to send.
That division of labor is what makes the workflow practical: one tool helps you
write faster, and the other helps you hand the result to someone else without
turning it into a formatting project.
Please enable JavaScript to use depapel.