How I Actually Use ChatGPT Projects to Run My Week, Instead of Starting Every Chat From Zero

For the best part of two years I used ChatGPT the same way most people do: one long, undifferentiated scroll of chats, each one starting with me re-explaining who I am, what I'm working on, and where I left off. I'd heard of Projects for months and dismissed it as a filing feature I didn't need. Then I actually set four of them up properly, and realised the filing was never the point — not repeating myself was.

I want to be honest about why I put this off, because I think most people skipping Projects right now have the same reason I did: it sounds like admin. Folders, labels, another thing to maintain on top of the actual work. What finally changed my mind wasn't a feature announcement, it was counting how many times in one week I'd pasted the same three paragraphs of context — my newsletter's voice, a client's brief, my own household budget — into a brand new chat because the old one had scrolled out of reach.

What a Project actually is, once you strip the marketing

A Project is a workspace, not a smarter version of ChatGPT. You create one, give it a name, attach whatever files it needs permanently, and write a short set of standing instructions once — tone, constraints, the things you'd otherwise repeat at the top of every chat. Every new conversation you start inside that Project can see those files and instructions automatically. Nothing about the model changes; what changes is that you stop being the one carrying the context around in your head.

That distinction matters because I originally expected something closer to a Custom GPT — a configured little tool with its own icon. It isn't that. It's closer to a labelled drawer that already has the right paperwork in it before you sit down.

The four Projects I actually run

I didn't try to organise my whole life into Projects on day one, which I think is where most people overreach and quietly abandon the whole idea within a fortnight. I picked four things that genuinely recur every single week, each with its own real reference material:

• Newsletter & Content — holds my voice guide, the last six issues, and a running list of topics I've already covered, so a draft doesn't repeat a joke I made in June.
• Client Briefs — one active client's brief, past deliverables, and their specific formatting preferences, so I'm not re-pasting the same style notes into every new draft.
• Household Admin — this month's budget spreadsheet and a running list of renewals and bills, which turned out to be the one I use most, oddly.
• Personal Research — whatever I'm quietly reading up on that week, kept separate from work entirely so it doesn't bleed into a client's context by accident.

"I didn't need ChatGPT to be smarter. I needed it to stop making me the human glue holding four unrelated conversations together in my own head."

What actually changed, day to day

The obvious win is speed — a new chat inside Household Admin already has this month's numbers loaded, so a question about whether I can afford something takes one line, not a re-upload. But the real change was quieter than that. I stopped dreading opening a new chat for recurring work, because a new chat no longer meant starting from nothing. It meant opening a drawer that was already half-organised.

The Newsletter Project is the one that surprised me most. Six months of past issues sitting there means a new draft genuinely reads more like me, faster, because it can check itself against what I've actually written before instead of guessing at my voice from a two-line instruction. It's a smaller, calmer version of the same instinct behind the automations that saved me 12 hours a week — the win isn't a flashier answer, it's removing a repeated chore from between me and the work.

Projects solved deciding what needs doing. They never solved when it actually happens — I still had to manually find the gap in my calendar for it, every time. That's a different job, and it's the one I handed to an app afterwards, rounded up in the only three AI scheduling apps I kept using.

Where it still falls over

I want to be straight that it isn't seamless. Projects don't talk to each other — a fact I only learned the hard way, halfway through asking my Household Admin Project a question that actually needed a client invoice sitting in a different Project entirely. There's no cross-referencing between them by design, which is mostly the right call for keeping work and personal life apart, but it means you occasionally have to consciously decide which drawer you're standing in front of. And a Project's usefulness is entirely proportional to how well you maintain the files inside it — mine went stale within a month until I built a five-minute habit of swapping in the newest brief or budget every Sunday.

It's also worth saying plainly that a Project is not a substitute for a proper system if the thing you're organising is genuinely complex. I still keep the actual structure of my week in Notion, not ChatGPT — the Project is where I go to think and draft with the right context loaded, not where I plan the week itself. Deliberately scoped context like this is also why I ended up turning off ChatGPT's global Memory feature once I had Projects doing the same job on my terms — the folder decides what's remembered, not a background feature quietly guessing.

Is it worth setting up

If you only ever open ChatGPT for one-off questions, no — a pinned chat does the same job with none of the setup. Where it earns its keep is anything recurring with real reference material attached: the same brief, the same voice, the same numbers, week after week. If that describes even one part of how you work, it's worth the ninety minutes. For a wider look at what else is actually worth setting up rather than just trying once and abandoning, the AI Tools hub is where I keep everything I've tested that's still in daily use months later, not just the week I wrote about it. Claude has the same feature under the same name, and I set mine up the same deliberate way for running this business, if you'd rather compare notes across both.

The same reusable-prompt trick does the heavy lifting elsewhere too — a saved Project is exactly how I keep the summary format identical in my meeting-notes-to-action-items workflow, meeting after meeting, feeding off whichever notetaker actually joins the call — I've only ever stuck with one of the three I tried. It's the same instinct behind the five ChatGPT prompts I now just paste instead of rewriting — a Project pins the files, a saved prompt pins the ask. If you want the next honest workflow write-up as soon as I've actually lived with something long enough to trust it, join the Brite newsletter.

If you want a fixed routine to feed that project, my twenty-minute Sunday review is the one I run it from: five short steps, with the AI used only to challenge the plan rather than write it.

Questions people actually ask

What's the actual difference between a ChatGPT Project and a Custom GPT?

A Project is a workspace: a folder of chats that all share the same pinned files and standing instructions, and you keep talking to plain ChatGPT inside it. A Custom GPT is closer to a separate configured tool with its own name and icon, usually built to hand to someone else or reuse across many different contexts. If the goal is keeping one ongoing area of your own life organised, a Project is the simpler fit; if you're packaging a repeatable task for other people, a Custom GPT is worth the extra setup.

Do ChatGPT Projects work on the free plan?

Projects themselves are available on the free tier, but the free plan's message and file-upload limits fill up fast once you're running several active Projects with attached documents. I'd only bother on a paid plan, because the whole point is keeping context loaded without watching a usage cap.

Can I share files inside a Project?

Yes, and it's the actual reason a Project earns its keep. You upload documents once to the Project itself, not to an individual chat, and every new chat inside that Project can reference them without you re-attaching anything. That's what stopped me hunting for the same brief or spreadsheet every time I opened a fresh conversation.

Does a Project remember things between separate chats inside it?

It remembers the files and the custom instructions you've set for that Project, and it can see other chats in the same Project if you point it at them, but it doesn't automatically fold every past conversation into a new one the way a single long chat thread would. Think of it as shared shelving, not a shared memory that reads your mind.

Is this actually better than just using folders or pinned chats?

For anything you touch once and forget, no — a pinned chat is fine. Projects earn their keep specifically on recurring work with reference material attached: the same brief, the same household budget, the same newsletter voice guide, week after week. The saving isn't the folder, it's never re-explaining the context that lives inside it.