Both ChatGPT and Claude added “Projects” — a way to group related chats, files, and instructions so a long-running piece of work stays in one place instead of scattered across dozens of conversations. I moved two real workflows (a client project and this blog) into each for a week. Here’s the honest comparison.
What Projects actually do
A Project bundles three things: persistent instructions (custom context the AI applies to every chat in it), reference files it can draw on, and all the related conversations in one folder. The promise: stop re-explaining context every time.
ChatGPT Projects — strengths
- Clean, simple folder model that’s easy to set up.
- Tight integration with the rest of ChatGPT (voice, web, image, tools) inside the project.
- Good for people who live in ChatGPT already.
Claude Projects — strengths
- Bigger, more reliable use of reference files — Claude’s large context means it leans on your uploaded docs well, which matters for context-heavy work.
- Strong at holding a consistent voice across a long project — useful for writing. (It’s why Claude wins my Claude Pro vs ChatGPT Plus pick for writers.)
- Project-level instructions feel like they actually “stick.”
The week-long verdict
For writing-heavy, document-grounded work (a book, a content site, research), Claude Projects kept things more coherent — it referenced my files more faithfully and held voice better. For a mixed workflow that also needs voice, images, and web inside the same space, ChatGPT Projects was more convenient because everything lives together.
The deciding question: is your project mostly text and documents (lean Claude) or a mix of media and tools (lean ChatGPT)?
One habit that matters more than the tool
Whichever you pick, write a tight project-instructions block — who you are, the goal, the voice, the constraints. That single paragraph does more for output quality than switching tools. It’s applied prompt engineering at the project level.
Related reading
- Claude Pro vs ChatGPT Plus: A Decision Tree
- The 10 Claude Prompts I Copy-Paste Every Week
- Every AI guide — the Library
About the author
Shahid Saleem is the founder and editor of PickGearLab. He tests AI tools in the real world — writing, automation, content — and writes up what actually worked. Based in Dubai.
One practical AI tutorial. Every Monday.
Workflows like this one — straight to your inbox. Free. Unsubscribe in one click.
Subscribe free →
