What to Do Before the Custom GPT You Paid For Disappears

How to preserve the useful parts of a GPT you’ve invested in, even if its creator doesn’t offer a replacement.

I’ve rewritten this around the reader’s investment, the approaching loss of access, and what they can do now. The creator’s proprietary information is addressed briefly, rather than becoming the focus.

What to Do Before the Custom GPT You Paid For Disappears

You bought a Custom GPT because it promised to help you do something.

Maybe it cost $27. Maybe it was $97. Or perhaps you’ve invested several hundred dollars in a collection of specialised GPTs.

You used them, learned how they worked, and gradually made them part of your regular business routine.

When you purchased them, you probably weren’t thinking about what would happen if the platform stopped supporting them. You expected to keep using what you’d paid for.

But with December 11 approaching, that expectation is becoming a real concern.

What if the creator doesn’t offer a replacement? What if they release a Skill version but ask you to purchase it again? What if the GPT you rely on simply disappears?

You’ve already invested money, time and effort into these tools. Understandably, you don’t want to lose all that value.

Fortunately, there are things you can do before access ends.

In a Nutshell

If you’ve purchased Custom GPTs, don’t wait until December 11 to find out what happens to them.

Check whether their creators are offering replacements, identify what makes your most valuable GPTs useful, and start preserving those working patterns in independent Skills.

You may not be able to recreate every feature exactly, but you can often retain much of the support you’ve come to depend on.

You paid for something you expected to keep using

A purchased GPT isn’t necessarily just another interesting AI tool.

It might be the one that helps you plan your marketing, work through an offer, write better emails or organise a complicated project.

You’ve learned what information to give it. You know which questions it asks. You may even have changed the way you work because of it.

As we explored in The Custom GPT You Built Your Business Around, some GPTs gradually become part of your working system without you even noticing.

Losing access means losing more than the original purchase price. It can also mean losing a familiar process that helps you get things done.

And if you’ve purchased a bundle of GPTs, you may be facing that possibility several times over.

You don’t necessarily need to recreate everything you’ve bought. But you do need to identify what you genuinely want to keep.

First, find out whether the creator has a replacement planned

Before doing any conversion work, check whether the original creator is providing a Skill version or another way to continue using the product. If you’ve bought any of my GPT assisted products you are covered. I will be updating the products with skill versions over the next couple of months.

Others might be offering existing customers a replacement, an upgrade, or a different arrangement.

If you haven’t heard anything, ask.

You could send something as simple as:

Hi [Name],

I purchased your [GPT/product name] and have been using it regularly.

With Custom GPTs disappearing on December 11, are you planning to provide a Skill version or another replacement for existing customers?

I’d love to know what options will be available.

Thanks!

If the creator is offering a replacement, find out whether it includes the features you rely on and whether there’s an additional cost.

You can then decide whether that option makes sense for you.

But don’t put off making your own plan while waiting for an answer. You still have an opportunity to work with the original GPT and learn from it.

What if there’s no replacement, or you don’t want to repurchase it?

This is where creating your own Skill becomes useful.

You may not have access to the original GPT’s instructions, reference files or internal setup. That doesn’t mean you have to start with a completely blank page.

You already know something important: what the GPT actually does for you.

Think about a purchased GPT that helps you write sales pages.

When you use it, perhaps it:

• Asks about your offer and intended customer.
• Helps clarify the main problem your offer solves.
• Asks about objections or concerns.
• Organises your answers into a sales-page structure.
• Checks whether important information is missing.
• Produces a draft you can work with.

You can observe all of that through ordinary use.

The first step is to document the experience, particularly the parts that make it more useful than simply asking ChatGPT to write a sales page.

For a practical introduction to this approach, see Can You Turn a Custom GPT into a Skill? Yes, but Start Here.

Pay attention to what makes this GPT worth keeping

Two GPTs can perform the same general task and still give you very different experiences.

One might immediately produce a long answer. Another might ask three important questions before it begins.

One might give you dozens of suggestions. Another might help you choose the two that are most relevant to your situation.

If you’ve paid for a GPT and find it consistently gives you useful results, those differences matter.

In Same Task, Different Value, we looked at why two tools doing the same job can provide very different value.

As you work with your purchased GPT, notice:

• Which questions does it ask?
• What information does it seem to prioritise?
• What order does it follow?
• Does it challenge an assumption or notice something missing?
• What makes its finished output particularly useful to you?

Don’t worry about identifying every detail.

Start with the behaviour you would genuinely miss.

What if the GPT relies on something you don’t have?

Some purchased GPTs use reference documents, examples, specialised methods or connected services that aren’t available to you.

You might be able to observe the GPT’s behaviour without having access to everything behind it.

That’s fine. You don’t need to crack open someone else’s GPT or retrieve their private instructions.

Instead, ask a more useful question:

What job was that missing resource helping the GPT perform?

For example:

The original GPT uses…	Your independent Skill could use…
A private brand reference	Your own brand and business context
A specialised evaluation checklist	Criteria developed for your needs
A collection of reference examples	Examples from your own business
A connected research service	A research tool you can access
A particular workflow	An independently developed process that accomplishes the job

The result may not be an exact replica, especially where the original depends on resources you can’t access.

But your goal is to retain the support that matters to you, not necessarily every component of the original product.

You might discover you don’t need every part of the original

Imagine you’ve purchased a GPT that produces twelve different kinds of marketing content.

In practice, you mainly use it to help create workshop sales pages.

Do you need to recreate all twelve functions?

Probably not.

You could begin with a Skill that does that one job particularly well.

This also gives you an opportunity to improve the experience.

Perhaps the original GPT always asks questions that aren’t relevant to your business. Your Skill could use your existing business context instead.

Perhaps it produces five options when you’d rather receive two, with a brief explanation of which situations each suits.

You can make those choices yourself.

That’s the approach behind Keep What Works. Then Make It Work Better for You.

Start by preserving the useful behaviour. Then adapt it to how you actually work.

Test your replacement while the original is still available

This is one of the most important things you can do before December 11.

Once you’ve created a first version of your Skill, give it the same task you would normally give the purchased GPT.

Use the same starting information, then compare the experiences.

For example, suppose your original sales-page GPT asks about likely customer objections before writing anything.

Your new Skill goes straight to producing a draft.

That’s a useful difference to notice. You can add the missing question to your Skill and test again.

Look beyond whether the finished documents sound similar. Compare how the two tools help you think and work.

In The Four Layers That Make a GPT Feel Genuinely Useful, we examine four things worth paying attention to:

1. Function: What job does it do?
2. Workflow: How does it guide you through that job?
3. Judgement: How does it decide what matters?
4. Distinctive Method: What particular approach makes it valuable?

You don’t have to get all four layers perfect on your first attempt. They simply give you a clearer way to recognise what your replacement might be missing.

And you have something valuable available right now: the original GPT to compare it against.

Three things worth saving before December 11

You don’t have to convert your entire GPT collection immediately.

Start with one you use regularly, particularly one that would be difficult to replace.

Then do three things.

1. Save a representative interaction

Give the GPT a familiar task and keep a record of the questions it asks, the information you provide and the result.

Use your own business information and examples.

Choose something that shows the GPT doing the job you value, rather than an unusual test that you would never normally run.

2. Note what you particularly appreciate

What does this GPT do that makes you return to it?

It might help you get started, ask questions you wouldn’t think of, keep you from wandering off-topic or organise information in a way that makes the next step obvious.

A few specific notes are much more useful than writing, “It’s really good at marketing.”

3. Create and test a small replacement

Build a first Skill around that useful behaviour.

Try it on the same task. Notice what’s missing, make an adjustment and try again.

You can develop the Skill further later. For now, you’re making sure that the most useful parts of the original don’t disappear without a trace.

What if you’ve created your own GPTs, too?

The process is similar, but you have an additional advantage.

If you created the GPT, you can generally work directly from the instructions, examples and supporting materials you own and still have access to.

You don’t need to reconstruct everything through observation.

For purchased third-party GPTs, your experience of using them becomes the starting point for developing your own independent solution.

Either way, begin with the GPT whose loss would make the biggest difference to your work.

You don’t have to recreate your entire collection

If you’ve accumulated a lot of Custom GPTs, the thought of converting all of them might feel like another enormous project.

It doesn’t need to be.

Some of those GPTs may be sitting in your collection because they looked interesting when you bought them. Others may have been useful once but aren’t part of your current work.

Focus on the ones that still earn their place.

Choose one you use, one you would genuinely miss, and one you can still access to test.

Preserve that value first.

Then decide what deserves your attention next.

Want to keep the GPTs you’ve already invested in?

The GPT to Skills Workshop will help you take a Custom GPT you value and begin turning its useful behaviour into an independent, portable Skill.

We’ll work through how to examine an existing GPT, identify what makes it useful, create your own version, and compare the results so you can improve it. Don’t worry, I have tools to help you do all that as painlessly as possible.

You can use this approach with purchased third-party GPTs or GPTs you’ve created yourself. Though for one’s you’ve created yourself there is an even simpler method.

You don’t need to convert your entire collection. Bring one GPT you don’t want to lose.

Keep it. Improve it. Take it with you.

Find out more about the GPT to Skills Workshop

I Have Questions

I’ve paid hundreds of dollars for a collection of GPTs. Do I have to convert all of them?

No. Check what their creators are offering, then prioritise the GPTs you actually depend on. You can decide what to do with the rest later.

Will my new Skill work exactly like the GPT I purchased?

Not necessarily. Some features may depend on resources you don’t have. Focus on preserving the behaviour and results that matter most to you, then refine your Skill through testing.

What if the creator wants me to purchase the Skill version again?

You can consider their offer, check the terms of your original purchase, and decide whether the replacement is worth the additional cost. Developing your own independent Skill may be another option.

What happens if I wait until December 11?

Don’t wait! According to ChatGPT, you will lose access to the Custom GPTs entirely (the ones you created and the ones you paid for). You will have no way to recreate them.

Does this work with free GPTs, too?

Yes. The same process applies whether you paid for the GPT or simply found one that became particularly useful to you.