What I want from a ChatGPT creative workspace for long books and reusable style

Here’s the gap I keep hitting. When a project runs longer than a week, ChatGPT becomes a conversation you have to re-explain every time you open it. I’m not asking for a new product — I’m asking for persistence so I can actually ask chat gpt to remember what I already told it.
I’m a machinist who writes books on the side. Over the last few months I’ve used ChatGPT to develop manuscripts, generate illustrations, edit text, and format pages for print. What started as a scratchpad turned into a full-time collaborator, but the moment I close the browser, the session ends and the context dies. That’s fine for quick questions. It breaks down hard when you’re twelve chapters into a novel or running a coloring book series with strict style rules.
The core problem isn’t capability — it’s continuity. Every new thread means re-uploading files, re-describing tone, re-defining character designs, and re-setting deadlines. By the time I get back to where I was, momentum is gone.
What I want is a permanent workspace tied to a project name. Not a folder. Not a thread. A persistent environment where everything lives: manuscripts, illustrations, character models, editorial standards, formatting specs, marketing assets, and a running history of decisions. When I open it, the AI already knows what I’m working on.
This isn’t theoretical. I’ve felt the friction every day. Without persistence, workflows collapse under their own repetition. A workspace would change that.

Permanent project libraries

Everything for Lost & Found StoryWorks would live in one place. No hunting through downloads. No re-uploading the same character sheet for the fifth time. The system would index files by name, chapter, or scene so I can say “open Chapter Nine” and it just works.

Persistent project standards

Right now I encode style rules into prompts manually. A workspace would store them — approved word lists, illustration dimensions, typography choices, print margins — as structured specs the AI references automatically. Instead of typing “use the same style as last time,” the AI already knows.

Direct document interaction

I should be able to point at a paragraph and say “tighten this,” or select a chapter and ask “find repetitive phrasing.” The AI would read the file in place, suggest edits, and apply changes without forcing me to copy-paste between windows.

Precise visual editing

When an illustration needs a tweak — change one character’s expression, adjust a doorway’s shape — I shouldn’t have to regenerate the whole image. Region-based editing keeps what works and fixes what doesn’t.

Reusable creative templates

For series work, a “Create New Page” button would load approved references, dimensions, and recurring elements. New scenes start from a foundation instead of zero.

Integrated production tools

The workspace would handle final checks: margins for print, image resolution, format compliance. No jumping between tools to ship.
The benefit compounds over time. The first week feels nice. By month three it’s essential. By year one it’s impossible to imagine going back.
I don’t need flashy features or real-time collaboration. I need the AI to remember what I already paid attention to. That’s the difference between a tool that helps and one that actually lets you finish.
I submitted this as feedback to OpenAI because I genuinely want to ask chat gpt to be a partner in long-term creative work — not just a very smart scratchpad that forgets everything when you close the tab.