Goodbye Custom GPTs. What's next?

[00:00:00] The easiest option out there right now could cost you the thing that you spent years building. Today, I'm going to walk you through exactly what to do with your custom GPTs, whether they live inside a paid program or you give them away for free. Welcome to the next evolution of the podcast, Entrepreneur School (In the AI Era). We are here to figure out how to integrate AI into the business

[00:00:25] you've worked so hard to build in a way that still feels like you. I'm your host, Kelly Sinclair, award-winning marketer turned AI platform co-founder. AI has fundamentally changed how I thought about work, systems, and what's actually possible. This show is about navigating that new reality together. It's going to be a wild ride, my friend, but I truly believe there's a way to leverage AI in a way

[00:00:52] that's intentional, human-centered, and aligned. It's an ongoing evolution, so let's explore, because AI may just be the unlock you need to achieve the life-first business you truly desire. If you build custom GPTs for your clients or your audience, you have got a decision to make before December 11th. And I want to make sure that you make it on purpose, because the easiest option out

[00:01:21] there right now could cost you the thing that you spent years building. Today, I'm going to walk you through exactly what to do with your custom GPTs, whether they live inside a paid program or you give them away for free. I'll also tell you where I'd draw the line on your IP and why I think a lot of people are about to hand theirs over without even realizing it. So custom GPTs are where I started with AI,

[00:01:48] and I bet a lot of you did too. They were one of the first ways that a regular business owner could build a tool for someone else to use, and that felt huge. You could take your process, you could write some instructions, upload a few files, hand it over to your clients, and they could start working through something that showed the way that you do the work you do. And I built a lot of them, and Brand

[00:02:13] Calibrator is my trademark custom process has been in custom GPT for years. But they also frustrated me because each GPT lives in its own little room, they couldn't talk to each other. So if your method had five steps and you built five GPTs, your client was the one copy and pasting between them, right? And they have no memory of what happened the last time that you engaged with it. And you have very little control over who got

[00:02:39] access for how long. So pros and cons of custom GPTs. And when Andrew and I built Wave, we kept the parts of custom GPTs that worked, and we fixed the parts that drove us nuts. So bots that pass context to each other, we built squads that would help follow a process from start to finish. We created a system

[00:03:02] that allows you to control access to who can actually use the tools that you make. So when OpenAI announced that they're retiring custom GPTs, I have feelings. Not surprised. But mostly, I just want to help you figure out what to do next. So here's a short version of the news. Back in August, OpenAI switched off

[00:03:30] new custom GPT building for personal accounts. And now it's official, they're retiring the custom GPTs completely. So their current FAQ, which I will link in the show notes, states that enterprise workspaces will lose theirs on December 11, 2026. And every other plan has been told to expect that same timeline. So two things I want you to hear. First, keep an eye on that timeline because it could shift.

[00:03:58] And second, ChatGPT is not emailing you about this. So if you don't go looking, you might not know. I found out while I was blissfully ignoring the world, probably on my third glass of Chianti sitting in Tuscany a few weeks ago. So no judgment if this is news to you. So here's how you start making your

[00:04:20] migration plan. Before you do anything, sort your GPTs into piles. Pile one is GPTs that you use only for yourself inside of your own business. So whatever that looks like, content workflow, admin stuff, internal operations. The good news is it's super easy. You can convert those into skills.

[00:04:44] Skills. You can use them in ChatGPT or in Quad. And quite honestly, they'll probably work better than they do now because skills run across your entire AI environment and they can pull from everything else that you've set up. They just get used when they need to be used, which is exactly what you want when the only person using them is you. So just copy those GPT instructions into a doc today. Like,

[00:05:12] go and organize all your GPTs. Take the instructions and the knowledge files and put them in folders that you have access to. Just in case you are like most of us, then you just like, they're just living in your ChatGPT account. And then you can rebuild them as skills whenever you're ready. All you need to do is upload those files and tell whichever LLM you're using to turn it into a skill. And that process will

[00:05:39] be very simple for you. Pile two and three are where it gets more interesting. So is this a free tool that you give away? Or is it a tool that you use inside of a paid program? And that question really changes what you should do next with it. So let's start with the paid ones because I think that's where most of the

[00:06:02] risk is for you. If your GPTs live inside of a membership, a course, a coaching program, or anything that people pay for, step one is going to be communication. You're going to need to talk to your clients about this change that's happening. Tell them that these current GPTs will stop working in December and tell them what's coming next, even if you haven't decided. And I'm on it and here's

[00:06:29] when you'll hear from me will go a long way. And you're going to need to make a bit of an internal transition plan around how you're sharing. Where are those links? Where did they live? How do we currently tell people about how to access the plots that we have? All of that just, it's going to be annoying. I'm sorry. I'm sorry it is. That's real. But step two is really thinking about intentionally about how you're going to rebuild. So here's my assumption. If your GPT is going to paying clients,

[00:06:59] it has your IP in it. So it's walking them through parts of your process or your methodology or framework, and that's the stuff that they're paying you for. So how you restructure it matters. And one of the most popular migration options right now is just turn your custom GPT into a skill file. I love skills. We talked about that. I've built over 60 of them at this point. They work across

[00:07:28] your whole AI environment. They solve that silo problem that custom GPTs had. And again, for running your own business, they are fantastic. But when you give a skill to someone else, they get everything that's inside of it. A skill file is basically a zip folder with your instructions, your reference files,

[00:07:51] every framework, every sequence, every judgment call that you have baked into it. The person who installs it can open that up and read the whole thing. So some of the creators that I work with at Wave have built their bot squads using full transcripts of their entire courses, that is the product. That

[00:08:14] is what people pay for. So I just can't even with why you would hand that over just because a platform forced you to change the format of the tool that you're using. And then there's the question of what happens when your IP goes into any AI platform that you don't control. There is very little clear governance right now on data collection and security in AI development. Even with that setting turned

[00:08:41] off, like where you click, do not let my stuff train the models, you're trusting a company with your life's work. And you really can't see what happens on the other side. So I'm seeing some conversation popping up. That's honestly driving me crazy. Like, AI has commoditized information. So IP doesn't really

[00:09:04] I don't hate that. I hate it. I really do. I think, yeah, information is everywhere. But your frameworks are not information. They're years of clients, mistakes, pattern recognition, judgment calls, the way you've learned to approach things that actually work. And we are the last generation of

[00:09:29] online business owners who built our IP before AI existed. We earned it without AI helping us. And I don't think we should have to give up ownership of that just because a tool changed. I feel like such a millennial right now, like the last generation of people to grow up without the internet. Yes, facts. And that means something, right? You get to decide where your expertise lives.

[00:09:58] So make that decision on purpose. And even if you're fine sharing your IP, there is actually a second problem with handing your clients a skill. And this actually affects the results that they will get with it. So I'm going to get nerdy for a minute on this one. When your client installs a skill,

[00:10:21] it doesn't run in a clean room, right? Like a custom GPT is a little container by itself that doesn't interact with anything else that's happening inside someone's AI environment. And that's actually a good thing from a client experience perspective. With a skill, it runs inside your client's entire AI environment. And you've no idea what's going on in there. It means it mixes with everything else,

[00:10:49] like two years of chat history, a handful of free skills they grab from other creators, and whatever their AI has decided is relevant memory about them. So it's like handing someone a recipe, but they cook it with different utensils. They swap in their own ingredients. They add a few flourishes just because. But you wrote the recipe, and they run the kitchen, and your name is still on the dish,

[00:11:19] if that analogy makes sense. Here's the nerdy thing. Here's the nerdy part. AI is what's called non-deterministic, which means basically you can't fully predict the outcome. If you ask it the same question on two different days, you're likely to get two different answers. And now you add a pile of context that you've never seen that's coming from someone else's AI environment. And you can't really

[00:11:46] test the outcomes and the outputs that are going to happen when they use it. Quick interruption, because if this conversation has you wondering what your expertise could become as an AI product, I want to help you build it. The AI Product Build Lab is a 90-minute hands-on workshop that includes

[00:12:09] a ready-to-install skill for Claude. It will help you map your expertise into a bot squad, identify what each bot needs to do, and then connect Claude directly to Wave so you can publish it as a paid AI product. You're not going to leave with another page of ideas to figure out later. You're actually going to build the thing. The regular price for this event is $97,

[00:12:36] but podcast listeners can join for just $1. So head to graviastudio.com slash build and use coupon code podcast1. That's graviastudio, G-R-A-V-I-A-S-T-U-D-I-O dot com forward slash B-U-I-L-D and use coupon code podcast1. Let's build your bot squad.

[00:13:16] So when I build bots and squads, I do it with three questions in mind. So the framework for building these is three parts. Number one, the input. What goes into it? What does the client need to bring to the conversation to make sure that your bots are actually performing and doing what they need to do? Number two is the process. So what's going to happen to that

[00:13:42] information? What frameworks is it going through? What research is it doing? What other reference material needs to be there to help create the output? And that's the third part, which is output. What comes out? What are you trying to achieve with the AI tool that you built? And in a contained environment, you can decide all three. Inside your client's AI,

[00:14:08] you write these instructions and essentially you're crossing your fingers that it's not going to get all mixed up with everything else that's happening there. So if the whole point of your tool is helping clients get better results in your programs, then that's really a lot of unknowns to put your name on. Something to think about and definitely something that's not come up in any of the conversation I have heard about this. Like it's like, what can we do to fix this? When the answer

[00:14:37] should be coming from the question of how do I get my clients the best results? Because to me, the real opportunity here is helping enhance your client's results with AI tools that are built with your frameworks. This expert backed AI that I've been talking about for over a year now,

[00:15:01] that's what it's all about. It's not about just continuing to have to pivot to what's available from the AI platforms that we're used to using. Okay, so let's move on to free GPTs because this is a bit of a different conversation. So start with a goal. Why does this free tool exist? And for most of us, I'm expecting it's for lead generation. Someone will use the free thing. They'll get a quick win.

[00:15:29] And they'll get a taste of how you think and then hopefully buy something else from you. So you're asking two questions. What is the purpose of this free gift? And then what is the best format to deliver the result that it creates? I'm assuming that you're not going to go back to PDF or a worksheet because that's where we all came from. And if you built a GPT, you wanted your free offer

[00:15:53] to feel like AI to actually implement the thing for them, right? So what are some of the AI options? There's definitely lots of options to consider. So I'm not really like going to tell you what to do, but just the things that you should think about when you're choosing what makes sense for your business because it really is a business decision. So you can build Claude artifacts,

[00:16:17] any kinds of interactive tools that you code with Claude's help. You could build an AI powered deliverable where maybe your client fills out a form or something that information runs through an AI process. It's connected with a tool like Make that does multi-step automations, and then it delivers

[00:16:40] sort of a mini report or something back to your client in their inbox. These are options. But it also depends on two things. Number one is how AI savvy you are. So if you heard me just going through that and you're like, that sounds like 150 steps that I need to learn how to do. True. And also how many other tools are you willing to pay for? The form to report setup might mean you need to pay for a form

[00:17:07] tool, an automation tool, and then also the AI credits that are used as part of that. And then, of course, email platform and all these things have to talk to each other. And you have to be there to maintain it if something breaks. So the other thing is, as far as format goes, if you found that having an interactive experience is really an effective way of getting your client to the result,

[00:17:37] that that's what a custom GPT was really good at, asking questions, getting follow-ups, working through something, and then your thinking is behind the scenes guiding them, then that's where I'll be straight with you. I haven't yet found another option besides Wave that can support this kind of thing. So yes, that's part of why we build it. And I want to share a little bit about how Wave

[00:18:02] could potentially help and also invite you to an event that we're having where you could come and test this out and see if this would work for you. So for your paid program GPTs, this is exactly what we build Wave for. Your GPT or multiple GPTs become a bot squad. Your method stays locked in. Your clients

[00:18:26] bring to the process what they need and everything else stays out. Bots can pass context to each other. So your client moves through your whole process in one place without them copying and pasting and moving back and forth between tools. And you keep your IP protected and you control access for how long and whether they're paying for it. And you control access to who can use it for how long. You just turn

[00:18:55] it on and off. For the free GPTs, if the conversation is the point, you can absolutely host these on Wave too. I currently have a free gift that lives there as a bot squad. So I have used Wave as a lead magnet for Wave because of course I did. But there's a platform cost, right? Every conversation that happens

[00:19:24] with your clients and your bot squad uses AI credits. And on Wave, you can choose whether you cover your users usage or they cover it. So with a free tool, usually you're the one covering it. So a free bot needs to really earn its keep by moving people towards something that they will pay for. So you have to kind of understand how often are they going to use this? What would that actually look

[00:19:49] like in terms of cost for you? And if you are also building paid products on Wave, then it might make sense because you are also using that same platform for something that people are giving you money for. And typically we see our clients paying for their AI usage, like paying for their clients' AI usage as part

[00:20:13] of what they're offering because the bot squad is part of a bigger program. It's part of something that maybe also includes access to you and whatnot. So you're just including it as one of the benefits of them taking your program or being in your world, your membership. So all that to say, if you have custom GPTs, they're not going to work after December 11th at the very least. And who knows

[00:20:42] if that will change? And you want to figure out what you could potentially do with them, then I want you to come to our next AI product build lab. We are hosting these events regularly. And if we're not near one when you hear this, there should be an ad on this episode that tells you when the next one is. And even if you can't come live, there's a replay, you'll immediately get the replay from the

[00:21:09] previous version of the event. But particularly at our next event, we will be talking about custom GPT migration, how it works on Wave, what that looks like, and gives you the opportunity to test it out and see if this is the right platform for you. So come to our next event. The link is in the show notes.

[00:21:32] It's graviastudio.com slash build. And you can use the coupon code podcast one to get the event for a dollar. So it's a very low barrier to entry to get yourself a two week free trial on Wave so you can play with it, test out your GPTs, see how it works, and see if that's the right solution for you. So

[00:21:57] here's what I want you to walk away with. Internal GPTs, no problem. Those can turn into skills in an afternoon. GPTs that you sell or give to people. So external facing custom GPTs. Make sure to communicate what's happening with anybody who has access to them. And to back up the instructions and knowledge

[00:22:24] files and everything that you have attached to that, you're going to want to make sure that you have your IP. So I hope that was a helpful guide for starting to think through your custom GPT migration strategy. I'm here if you want to talk it through, you can reach out to me directly via email kelly at graviastudio.com. We're happy to jump on calls with you show you how Wave works. And I will be completely honest if I think that Wave is not the right fit because it's not so the right

[00:22:54] fit for everybody. So it just happens to be a nice opportunity for us to talk about the platform that we're building, given the previous version of the tools and custom GPTs are. They're going bye-bye and thanks OpenAI, I guess. All right. We'll talk to you next week. Thank you for listening. If this episode got you thinking about how AI is impacting your business

[00:23:20] and what it might look like to integrate it in a way that actually fits, I'd love to help you think that through. You can book an AI strategy chat using the link in the show notes. It's a space to talk about where you are, what's shifting and how to move forward with intention. And if this episode was helpful, please share it with another business owner who's navigating this evolution too. And make sure you're subscribed so you don't miss what's coming next.