Keep what works. Then make it work better for you.

You may have collected a whole little library of useful Custom GPTs.

Some you made yourself.

But many of the really useful ones may have been created by other people.

You found one for planning. One for writing. One for analysing something. One that asked surprisingly good questions. One that saved you from staring at a blank screen.

And now the obvious worry is:

What happens when those GPTs disappear?

And they will on December 11, 2026.

The goal isn’t to recreate someone else’s GPT line by line. It’s to understand what made that GPT useful to you, then carry that value forward into a Skill you can keep using.

And yes, the same process also works for GPTs you created yourself.

In a Nutshell

You don’t need to own a Custom GPT for it to have become valuable to the way you work.

Maybe you’ve gathered GPTs created by other people and built them into your regular business routines. If one disappears, what matters most is not its name or even its exact setup.

What matters is the value you were getting from it.

When you understand that value, you can start rebuilding the useful experience in a Skill of your own.

If the GPT is one you created yourself, the process can go even deeper because you already know more about how it was built.

But either way, the first question is the same:

What would I actually miss?

The job is only part of the value

Imagine you use somebody else’s GPT to help turn rough ideas into blog posts.

At first glance, its job seems obvious:

Idea in → blog post out.

But another AI could perform that exact same task and give you something completely different.

One might immediately produce 1,500 words.

Another might ask you three questions first.

One might give you polished marketing copy filled with phrases you’d never say.

Another might help you untangle the idea before it starts writing.

Same task. Very different experience.

That’s why simply identifying what the GPT does isn’t enough.

You also want to notice what makes its way of doing it useful to you.

That applies whether you built the GPT yourself or found it in someone else’s collection, course, membership, resource library or shared link.

Look at the parts you’d actually miss

A good place to start is surprisingly simple.

Choose one GPT you use regularly and ask:

If this GPT disappeared tomorrow, what would I miss about it?

Not just what job would suddenly need doing.

What would you miss?

Perhaps:

• the questions it asks before it starts
• the order it does things in
• the way it turns vague thoughts into something concrete
• a particular type of result it reliably produces
• the amount of detail it gives you
• how it handles choices
• the things it avoids doing
• the way it helps you think through the task

Those details can easily look insignificant.

They’re often where much of the value lives.

And you can notice those things even when you didn’t create the original GPT.

You don’t need someone else’s hidden setup to learn from the experience

If somebody else created the GPT, their private instructions and proprietary material are theirs. You don’t need to uncover those.

What you can work with is your own legitimate experience of using the GPT.

You can notice:

• What did I give it?
• What did it ask me?
• What did it help me think through?
• What kind of result did I get?
• What did I like about that result?
• What did I repeatedly correct?
• What would I want my own version to do differently?

That’s enough to uncover a surprising amount of useful information. You’re not trying to clone someone else’s GPT. You’re figuring out what value it created for you so you can build a Skill that supports the same kind of work.

Don’t preserve annoyances just because they’re familiar

There’s another advantage to doing this.

The GPT you’ve been using probably isn’t perfect. Maybe you’ve learned that you always need to say, “No, not like that.” Perhaps you routinely delete the same unnecessary section. Maybe it gives you five alternatives when you really want one good recommendation. Or you’ve developed a strange little ritual where you give it the same correction every time you use it.

We get used to these things.

Eventually they become part of “how the GPT works,” even though they’re not things we particularly want.

When you build your own Skill, you don’t have to preserve every quirk.

Keep the value. Question the friction.

Your new Skill doesn’t have to be a copy

If you’ve been using somebody else’s GPT, your goal isn’t to reproduce it exactly. Your goal is to create something that works for you.

That may mean your Skill eventually:

• asks different questions
• uses a slightly different workflow
• produces a simpler final format
• includes your own preferences
• stops doing something that constantly irritates you
• connects better with the rest of your business
• becomes easier to update as your work changes

You’re using the original GPT as evidence of what’s useful, not as a mould you have to copy.

That gives you room to make the new Skill a better fit.

What if it’s one of your own GPTs?

The process still works. In fact, you have an advantage. If you built the GPT yourself, you can look not only at your experience of using it, but also at the instructions, files, examples and decisions you put into it. That can help you preserve more of the original logic.

But I wouldn’t start there.

I’d still start with the same question:

What would I actually miss if this disappeared?

Because even with your own GPTs, not everything deserves to be carried forward. Some instructions may be outdated. Some decisions may no longer fit your business. Some parts may have seemed important when you built the GPT but barely matter in practice.

The goal is still the same:
Keep what works. Then make it work better for you.

Preservation first. Improvement second.

It’s tempting to improve everything at once. You start looking at a GPT you use and suddenly think:

While I’m here, I could change this… and add that… and completely rebuild this other bit…

And before long, you’re no longer preserving something useful. You’re designing an entirely new system. That can make it surprisingly hard to tell whether you successfully kept the original value.

A better mental model is:

First, understand what you’re trying to keep.

Then, decide what would make it better.

Those are two different jobs. Keeping them separate makes both much easier.

You won’t need the original GPT forever

At first, losing access to somebody else’s GPT can feel like losing the solution itself. But often, what you really need is the result of the thinking behind your experience with it.

You need to understand:

• Why did I keep coming back to this?
• What problem was it solving?
• What made it easier than doing the task another way?
• What parts actually mattered?

Once you know that, you’re no longer completely dependent on the original GPT. You can begin creating something that belongs in your own system.

And if the GPT was one you created yourself, you can do the same thing with even more information available to you.

This is bigger than replacing a disappearing tool

It would be easy to look at the disappearance of Custom GPTs as a technical problem.

A tool you use is going away, so you need another tool. But if you’ve collected useful GPTs from other people, there’s more at stake than that.

You’ve also collected ways of getting work done.

Some of those GPTs may have become little shortcuts in your business. Some may help you think. Some may save you from getting stuck. Some may turn a task you avoid into something you can actually finish.

That’s worth identifying.

You’re not just asking: “How do I replace this GPT?”

You’re asking: “What was this GPT doing for me that I want to keep?”

That’s a much better starting point.

A small thing to do now

Choose one Custom GPT created by someone else that you use regularly.

Use it once with a familiar task, but this time pay attention to the experience, not just the answer.

Notice:

• What did it do that helped?
• What did it ask you?
• What did you have to correct?
• What would you want your own version to keep?
• What would you happily change?

If you’ve also created GPTs yourself, you can repeat the same exercise with one of those too.

You don’t need to rebuild anything yet. For now, just identify the value. That gives you something far more useful than a list of features. It gives you a picture of what you’d actually want to carry forward.

Your next step

In my GPTs into Skills Workshop, we’ll be working with the Custom GPTs you’ve already gathered and use.

They can be GPTs created by other people.

They can also be GPTs you created yourself.

The goal isn’t to copy someone else’s private instructions.

It’s to identify the value you’re getting from the GPT, understand what makes it useful to you, and begin turning that into a flexible Skill you can keep using.

You don’t have to start from scratch.

And you don’t have to lose everything useful just because the original GPT disappears.

The better question is:

“What am I getting from this GPT that’s worth keeping?”

I Have Questions

Can I do this with a Custom GPT someone else created?

Yes.

You can work from your own legitimate experience of using the GPT: what it helps you do, what it asks, what kind of output it produces, what works well for you and what you’d change.

You don’t need access to the creator’s private instructions to identify the value you want to preserve.

Does this also work with GPTs I created myself?

Yes.

With your own GPTs, you have even more information available because you can also review the instructions, knowledge files and other material you used to build them.

But it still helps to start by identifying which parts actually matter in practice.

Am I copying someone else’s GPT?

That’s not the goal.

The aim is to understand what you value about your experience using the GPT and create your own Skill for your own workflow.

That’s very different from trying to extract or reproduce someone else’s hidden instructions or proprietary material.

What if I can’t explain why a GPT works so well?

That’s common.

Use it with a task you know well and observe what happens. Notice what it asks, what choices it seems to make, what kind of result you get and where you rarely need to correct it. And let help ChatGPT help you.Please

You don’t need to understand everything at once.

Do all of the GPTs I’ve collected need to become Skills?

Probably not.

Some may do simple jobs that are easy to replace.

Others may have become an important part of how you work.

Start with the ones whose disappearance would create a real gap.