
What happens to my custom GPTs if the platform changes?
Short answer
They change or disappear with it, because work built only inside one company's app is rented. Make each assistant portable: copy its instructions and settings into a plain text file, in a folder on a drive you own, named for the job it does. Then your library survives any platform change.
A few weeks ago I opened one of my custom AI assistants and the menu had moved. The tool still worked. What stopped me was a smaller, colder thought: I had built dozens of these, and I did not actually own a single one of them.
Everything I had made lived inside one company's product. If they changed a feature, raised the price, or shut the whole thing down on a Tuesday, my work left with them. I had been renting the entire time and calling it building.
When does an AI tool count as yours?
An AI asset only counts as yours if you can move it. If it exists only where a platform lets it exist, what you have is a subscription with good manners.
The test is simple. Count the AI tools you have built inside one app, then ask where they go the day that app changes its price or kills the feature. If the answer is "nowhere," you have been renting your own work.
What did moving 56 assistants involve?
I had built 56 custom assistants over a couple of years, each one a little system for a recurring job, like a client intake or a research routine. I spent a day pulling all 56 out of the app and rewriting each as a plain, portable file in a folder I control.
Then I built one simple page that checks every tool still runs. Now if a platform changes its terms, my library walks out the door with me, intact.
The check page matters as much as the files. A folder of instructions you never test can go stale just like the originals did.
Portability does not mean leaving the platform. You can keep using the app. The difference is that the app holds a copy of work you keep elsewhere, so a change on their side costs you a re-upload instead of your library.
A day sounds like a lot until you compare it to the alternative: rebuilding 56 tools from memory after a change you did not choose.
Why does portability matter now?
AI products change their pricing and their rules often. The operator who keeps their best work in plain, portable formats holds on to it through every one of those changes. The one who leaves it locked inside an app is a single product update away from rebuilding from memory.
Do this today
Ten minutes, one tool.
- Pick the one AI tool or custom setup you lean on most.
- Copy its instructions and settings into a plain text file, in a folder on a drive you own, named for the job it does.
That is it. You now own one. The library is that move, repeated on the things you would hate to lose.
Meredith's rule
The platform is borrowed. The work does not have to be.
Questions
How do I back up my custom GPTs?
Copy each assistant's instructions and settings into a plain text file, store it in a folder on a drive you own, and name the file for the job it does. Start with the one you rely on most, then repeat for any tool you would hate to lose.
Is it risky to build all my AI tools inside one platform?
Yes. If the company changes a feature, raises the price, or retires the product, the work you built there can leave with it. Keeping a portable copy of every important assistant lets you rebuild it elsewhere instead of starting from memory.
How can I tell whether my AI tools still work after a platform update?
Keep a simple check page or list that runs each tool and confirms it still produces the output you expect. Pair it with portable text copies of every assistant, so a broken tool can be restored or moved instead of rebuilt from scratch.
Related service
AI Architecture Review

Want this worked out for your business?
Bring the real problem to a working session and leave with a plan.
Book a session