Canvas is a collaborative editing space where you and ChatGPT can create, edit, and refine content together in real time. Instead of bouncing between chat and a separate doc, Canvas keeps everything in one focused workspace — making it easy to co-write documents, structure projects, and even debug code.
Why it’s useful at work
Collaborate in real time
Work directly alongside ChatGPT in a live document. Edit text, provide in-line feedback, or highlight passages for AI-powered suggestions.
Streamline writing and coding
Draft announcements, reports, proposals, or code without the back-and-forth of copy/paste.
Boost quality and speed
Use built-in tools like Suggest edits, Adjust tone/length, or Polish to instantly improve clarity and style.
Code smarter
Open a code canvas to write or refine code, run snippets in the built-in console, and debug with AI assistance.
Getting started
Start a canvas – Type /canvas in chat or click the ➕ icon and select Canvas.
Add content – Paste in notes, transcripts, drafts, or code as your starting point.
Edit together – Highlight sections and ask ChatGPT to rewrite, expand, simplify, or refactor. Summarize this into key takeaways for executives” or “Rewrite in a friendlier tone.”
Polish & finalize – Use Canvas tools for tone, length, polish, or formatting. Export, share, or continue iteration.
Tips for working in Canvas
Define your goal at the top of the canvas so ChatGPT has context.
Don’t hesitate to iterate — small inline edits build up to a polished draft.
Use version history to restore earlier drafts when exploring different directions.
Examples for your role
Role
Use case
Ready-to-use prompt
Marketing
Turn a webinar transcript into a first-draft ebook
Draft an ebook outline based on this transcript, with 5 chapters and a short intro.
Product
Prototype from requirements doc
Create a simple HTML/React prototype based on this PRD.
Sales
Personalize outreach emails
Rewrite this outreach email for a healthcare buyer, making it warm and consultative.
Engineering
Debug Python code
Fix this error in my Python script and explain the issue in plain language.
