S1. Ep38 - How to Create Custom GPTs for Your Business
Wondering how to get consistent results from ChatGPT without typing the same instructions over and over?
You're not alone. In this episode, Katie and Noel break down custom GPTs, what they are, why they matter for business owners, and how you can build your own AI assistant without writing a single line of code.
They reveal how custom GPTs can save you hours every week by eliminating repetitive tasks, why the $20 ChatGPT plan gives you access to build unlimited custom GPTs, and how you can train them on your brand voice, FAQs, or any documents you need. Plus, they share real examples including a Make.com automation assistant that walks you through building scenarios step by step, a brand voice analyser that creates comprehensive voice guides from your content, and an ideal client profiler that digs deep into emotional pain points most people miss.
Perfect for business owners, coaches, consultants, and membership creators ready to scale without adding more hours to their day or team members to their payroll.
How to find us:
Join our membership over on Skool, where we support you on your AI and automation journey. We share exclusive content in the membership that shows you the automations we talks about in action how to build them. Find out more about the AI Business Club here.
We have a free LinkedIn group (AI Automations For Business), the group is open to all.
If you would like dedicated help with your automations or would like us to build them for you then you can find our agency at makeautomations.ai
-
Katie (00:26)
Hello and welcome back to another episode. Hi, hello, I'm Katie and as always I've got Noel here with me. Hi Noel.
Noel (00:35)
Hello, good to be here. I'm looking forward to diving into today's topic.
Katie (00:40)
I thought you might be. So today we are talking all about custom GPTs, what they are, why they matter for business owners. And like, I feel Noel, you've been playing around with custom GPTs for quite some time. I feel like you must because I feel like I've known about them for quite a long time. So you've obviously been in my ear talking about them.
Noel (01:08)
Yeah, I was building them as soon as they released them, which was the end of 2023. Yeah, it's been a while. I've created quite a few during that time.
Katie (01:17)
Yeah, it's been a couple of years. Yeah, okay. So like I said, I feel quite excited about this because I feel like there is so much untapped potential for business owners with a custom GPT.
Noel (01:34)
Yes, definitely. Yeah. You can, it makes it a lot easier. So yes, you could automate with tools. Whereas I think with this is really beginner friendly, nice and easy to do.
Katie (01:44)
Yeah, so let's just start like from the very beginning. So someone might be listening to this. I mean, like, what even is a custom GPT? So Noel, can you break it down for us really simply? What is a custom GPT?
Noel (02:03)
So what a custom GPT is, it's a way in which you can train. Basically, you're training your own little agent, and you can train it on anything that you wish. So it could be for any use case. You can give it specific instructions and then say, well, look, I need this particular output every single time. So you can train it on your documents. You can get it to do web searches. You can do it to create images. There's all kinds of different things you can do. It's kind of like a real cheat sheet to get to the end goal on ChatGPT. So if you have something really specific, then your custom GPTs are the way out.
Katie (02:48)
Yeah, I feel like a custom GPT almost acts like another version of you or even like a digital team member. And what I love is that you can actually do this without coding, because I know there are a couple of different ways that you can do it.
Noel (03:08)
Yes, there are two ways to do it. So first of all, to get access to custom GPTs, you do need a paid ChatGPT account. So you don't need to be on the super duper $200 plan, as long as you're on the, I think it's the core plan for $20. As long as you're on there, you should all be good to go. But yeah, to create one, they've got like a GPT explore area. And then right in the top right hand corner of that page is a little create button. And once you're in there, there's two options to build them. You could chat to, technically chat to GPT, to help you create the custom GPT you want. So you can just, you know, type it all out in plain language and describe what you want this GPT to do. Or you could go a bit deeper and then there's a configure tab where you can, you know, go through and you can define the system prompt. You can then add all the files that you need access to. You can add and remove different services. So if you don't want it going off and searching the web, you can, you know, untick that option. And there's all kinds of things you can link it into APIs and it gets very technical and that's on the configure tab, but you can still, you know, you can switch between the two. So let's say you start creating one in plain language and you're like, well, I'm not quite sure this is a hundred percent right. You can then go into the configure tab and then see what it's done. And then make those edits as well. So yeah, really easy to get started.
Katie (04:42)
Okay, so like, why do people like want to create one? Maybe someone's still on the fence and they're like, why go through this effort of like creating one? So for me, I feel like it saves a lot of time.
Noel (04:52)
Yeah. Definitely loads and loads of time. So if you're, like every week or every day, you're asking ChatGPT to do something for you, whether it be creating content or answering certain questions. And every single time you're like, well, this isn't giving me a consistent output. Because every time you could ask the same question six times and get six different answers. But if you're running into that sort of issue, then maybe a custom GPT is the way forward because you can go in there and say, well, look, this is the output format that I want. This is an example of something good that I want to receive back. And then it goes, sure, OK. You can just go off and ask those custom GPT those questions and you get it. I'd say 90% right every time. Let's not go with 100%.
Katie (06:00)
Yeah. Yeah.
Noel (06:03)
It can give us stress sometimes.
Katie (06:04)
Yeah. Because I would say the consistency is definitely all over the place if you're just using ChatGPT, whereas I feel like the consistency if you're using a custom GPT is a lot higher.
Noel (06:11)
Yes. Way higher. Yeah. It really gets into those instructions. So if you've really embedded that within the prompt, within the GPT, then yeah, you're just going to get it in that structure every time. Obviously it works two ways or you could have it so it's less structured. If that's the way you want to roll, you know, that's fine. But it's still got access to the information that you've given it as well. So, you know, it's still accurate and gives you what you need.
Katie (06:47)
Yeah, you're not having to put in that same information every time you're using ChatGPT.
Noel (06:54)
Exactly. Yeah. It'd be like, here's my PDF that's about my brand voice. And then you go now create me the content. Whereas if you create a GPT that has that document embedded in it, you can just go to it and say, create me the content about, you know, whatever you want to talk about. And it will go off and use that document to create it and things like that. So yeah.
Katie (07:00)
Yeah. And I know from the work that we've done for other people, it improves client experience as well.
Noel (07:27)
Definitely. Yeah. You can give them, I mean, you know, I guess for some clients, you know, you could give them access to information that's maybe like FAQs and things like that, you know, so they can get instant support from a custom GPT instead of, you know, sat around waiting for you to answer an email or, you know, or maybe not getting the answer they need from your chatbot on the website or whatever, you know, they can just get that information straight away. Yeah. They're really helpful.
Katie (07:58)
Well, they're kind of like not sat there waiting for you to come online in like maybe like a membership area, you know. Yeah, because I know, like, we'll probably talk about this in a moment. I mean, let's go into it right now, I guess. But for example, Noel has built before chatbots for memberships. So if a member has a question or doesn't know where to find something within the membership area, he built this chatbot that was actually quite fun and people could just put their question or what they needed help with into this chatbot and it would come back and find the answer.
Noel (08:57)
Absolutely. Yeah. I had a real fun persona. So that's also something you can bake into your custom GPT. So if you don't like the same boring, you know, AI response, you can add that in there as well. So, you know, it makes it a bit more fun and interactive. But yeah, for that chatbot I also created a custom GPT. So when we was interacting with people in comments and things in the community feed, we could interact with them and use that same persona. So we could put their comment or post into the custom GPT and it would come back with a response using that exact same persona that they would get from the chatbot within the platform. So yeah, that was quite fun.
Katie (09:42)
And I think this is great as well because obviously I've done community management for a couple of different memberships. And even though people know you aren't the person who runs the business, you know, they've joined because they like the tone and the vibe of that person. So you have to be very careful like how you are responding to members.
So very often you get trained in their tone of voice. You know, replying to any customer support emails or replying to any questions, comments in the community, you have to be kind of like, you know, is this how they would respond? So this like custom, you know, chatbot that you built allowed that to happen with, you know, without then having to spend time training members or staff and you know the worst thing as well is when you have like a really great member of staff and for whatever reason you know they leave you then have to like start all over again and people have to you know know the knowledge of the membership and understand the tone of voice and things like that all over again.
Noel (11:02)
Yeah, exactly. That's the hard bit. With this, you could share that GPT with the entire team. So you've got your company voice is in that GPT. You could share that with everyone and everyone gets the same outputs, it's going to be far better than, you know, 10 people firing up 10 instances of ChatGPT, asking questions differently and then hoping they're going to get a similar answer. Yeah. That's not really going to happen.
Katie (11:33)
Yeah. So let's go into some like, well, while we're here talking about it, let's go into like some other real use case studies for like business owners with custom GPTs. So, something that I have used myself is like a course curriculum builder. So, you know, you can create your GPT to kind of like, just help you map out a course so you can like tell it, like, I want to create a course on this subject, it's going to be, I don't know, six weeks long, or it needs to have at least two hours with the training, can you like build me the curriculum? And then I think sometimes when you've just got like the bones of the course, as it were, like the outline, I feel like then it's so much easier to go and write the course yourself. So you're not asking, you know, ChatGPT to write your whole course for you. You're just asking for like an outline to kind of get you started.
Noel (12:44)
Yeah, yeah, I've done similar things myself for the courses I've created because it's one thing having the idea about what you're going to train people on. But then yeah, it's for me, it's quite another thing to figure out. Well, which sections does it logically make sense to train this in? You know, how should I break all this information down? Yeah.
Katie (13:03)
Yeah, that's the thing. Like when you've got so much knowledge in one particular topic, it's kind of like, well, yeah, sometimes it's really hard to go like, right, well, how do I just, you know, instead of making this, you know, a three year course, how can I just, you know, zone in and concentrate on one bit? So it's only like an hour or two hours. And also I think as well, like you were saying, it's really good because it will help you put things in like a logical order. Whereas like in the past before I've gone to create courses and I've gone, right, well, I want them to learn this. And I'm like, but actually in order them to know this, you know, there's a bit that needs to come before that. And now actually does the order make sense? So even something like that can save you a lot of time. And I think as well it takes away that like writer's block as well, doesn't it?
Noel (14:11)
It does. Yeah. It's definitely helped me a hell of a lot. But once you've created that output, so let's say, you know, the first time you run it, your curriculum might come out slightly different, maybe not in the right sort of way. And then, you know, you might give it another go and be perfect, or you could then copy that perfect one back into the custom GPT and then say, well, look, this is a really good one. And then the next time you go to create a course, it will be in that exact layout, you know, and follow that same format. So all of your training that you do follows that same consistent style as well. Yeah, pretty powerful.
Katie (14:49)
Yeah. So you can even have like custom GPTs that, you know, you can give them like team or like job roles, can't you?
Noel (15:02)
Yeah, yeah, it's really quite fun because you could basically create like a job specification. I mean, you'll get ChatGPT to create that or find one online for like a specific role. Say if you want like a chief technical officer, you know, for your business, you know, you could pre-program a custom GPT that will act as your CTO so you can ask it all kinds of questions and have that back and forth like you would do with a normal employee. So it's like a solopreneur, you've kind of got to wear all of those different hats, whereas now you can put those hats into a custom GPT and then just sit there chatting to them all, which is really quite cool use case for them. And then also you can give them a fun persona as well. They don't have to be boring. They could have a bit of flair and style in there.
Katie (15:39)
Yeah, like you could do almost like you can give them like celebrity personalities as well, can't you?
Noel (16:02)
Yes. Yeah, I was creating an agent with the persona of Ron Burgundy from the Anchorman films. That was yeah.
Katie (16:09)
Wow. How was that working with that sort of personality?
Noel (16:14)
It was actually quite good, it was brilliant. Yeah, it was really fun. In the end, yeah, you didn't really get any work done, but yeah.
Katie (16:18)
Yeah, so probably not the best person to base it on. But it goes to show, yeah, but it goes to show, like, if you've got a favourite celebrity or film star, yeah, I would quite like to see what Reese Witherspoon would come out like.
Noel (16:27)
Yeah, fun but not too fun. Yeah, yeah definitely that'd be good. Or who else, like Ryan Reynolds giving you no nonsense advice, that'd be quite good.
Katie (16:47)
Yeah, because I feel... Well, I was thinking like Kristen Wiig or Amy Poehler, Tina Fey. Like they're all obviously really funny comedians that do a lot for like SNL. So I just feel like, yeah, but then maybe it goes back to like the Ron Burgundy, like how much work are you actually going to get done? Yeah. I think you need to be careful when choosing the personality for your custom GPTs.
Noel (17:06)
Exactly, yeah. Yeah, we were sat down having a virtual scotch and clinking glasses at one point. I thought that was the point, I thought this has gone a bit too far.
Katie (17:28)
Brilliant. Yeah, maybe. But no, you've also created some custom GPTs inside of AI Business Club, where you're guiding members through step by step to build brand voice, ideal client, and you've also got a Make.com assistant in there. Do you want to just tell us a little bit more about those ones that you've created.
Noel (18:06)
Yeah, so the brand voice, that's quite an easy one, actually. When our members use that GPT, all they need to do is like input some text examples of their content, their written content, obviously written by them, not by AI. So they can put that either as like a document or they could copy paste it into the chat. And then all they got to do is hit enter and away it goes and creates a really comprehensive brand voice, which you can then use to, you know, you could then copy that out and then put them into your own custom GPT, you know, which is trained on that information. So it's a bit circular, but it gets you started for creating a brand voice if you've never created one before. So yeah, really, really cool. That one is, the other one with the ideal client, you're kind of like describing your business and then what sort of services you want to provide and who you're looking to target. And then it goes into quite a bit of detail about these people, you know, not quite down to shoe size, but it's almost to that point, you know, it goes into so much information that it's like, well, you know, this is what they like doing, you know, it gives them like names, ages, locations, you know, all of the interesting stuff that you might want to know and to understand your ideal clients and things like that. So all of that information comes out. Again, that's a really simple text prompt to get that started.
Katie (19:36)
Yeah, and it also goes into like the emotional side as well of your ideal client as well. Yeah, what they've tried. Yeah, exactly. Because I feel like that is something that a lot of people actually miss out on like as well with their ideal client. And like, I think people forget like how much people buy based on emotions. So yeah, just having that in there, I know, you know, has helped people.
Noel (19:43)
Yes, yeah, what they're struggling with and yeah, it's all sorts.
Noel (20:10)
Yeah, absolutely. Yeah. The third one is the Make.com assistant. So I don't know if you've noticed recently, they seem to have moved their help AI in the builder. They seem to have changed that. So I'm guessing they're getting ready for Maia at some point, which is going to, you know, take that to the next level. But the GPT that I've built is kind of, I would say a Maia light. So you could describe to it and say, well, you know, I want to take emails from my Gmail account and then I want to then have AI that will go through, maybe do like some tasks to say, well, you know, is this like a sales email? Is this, you know, categorise it and all that sort of stuff and then generate responses. And what the assistant does is it goes, right, okay, well, I'm going to go off and find out about Make. And then it comes back and then gives you like a step by step guide on how to build the automation. So you go right, well, to start off with, you need the Gmail module and it will tell you what information to put in and then how to connect it into OpenAI or whichever AI you want to use. And then it would then be like, right, well, you need to link this data in this field. And it takes you through that step by step process. It might not be 100% perfect. Everything that I've tested so far it's got right and yeah it's been really helpful for some of our members.
Katie (21:40)
That's good. And obviously you had to create one for Make.com now, obviously.
Noel (21:48)
Of course I did, yeah. I had to shoehorn them into the episode somehow, didn't I?
Katie (21:53)
So, yeah, so I think when you're creating your custom GPT, like you have to think like what problems do you want to solve? What tasks do you want it to help you with? Your tone, give it a personality. It can be your personality. It could be, you know, something, someone unhinged. What information can you upload to train it if that's something that is required from it? And then don't forget to test and refine it and tweak if needs be. Yeah, when should you be building a custom GPT?
Noel (22:36)
It's when you're doing stuff repetitively, you know, at least, you know, once a week, you know, like for us, you know, we've got one that helps us with our podcasts. So every week I just nip in there with the transcript and then go right now, create, you know, the ideas for the titles and descriptions and that sort of stuff and the show notes. So that's something we do every single week. So that's something that's, you know, really good to tackle. Yes, whenever you're doing anything specific, whether it always needs to be repeatable, or you always need something out in the same format. And you know, you don't really want to go through the pain of describing the output format every single time that can get very annoying. So yeah.
Katie (23:33)
Yeah, I'm gonna add as well to the list, like when you want to provide like a seamless customer service, I'm gonna add that one in. And also when you want to be able to have more clients or customers, but you don't actually have any more time or you don't want to give any more time to it.
Noel (24:05)
Yeah, well, it's like sharing things within a team. So if you want your team to be using the same sort of outputs or using the same information, then having them having a consistent experience as well. So they're really good points for creating a custom GPT.
Katie (24:25)
Yeah. So I think we've given like custom GPTs like good PR here.
Noel (24:35)
Definitely, yeah. I think they deserve it. They are really good. Really, really good. Yeah. I mean, Claude released skills the other week. That's kind of edging into this sort of space, but it is slightly different and a bit harder to do. So I think, yeah, GPTs are still leading the way when it comes to creating this sort of thing.
Katie (24:36)
Yeah, I think they're great. Yeah. And of course, like people don't have to figure this out alone. You know, we've created some in AI Business Club. We also guide members, you know, step by step how to build their own. For example, you know, as well we've done like the ideal client one, the brand voice, Make.com assistant as well. So yeah, if you do want some hand holding, creating your own custom GPT, then we can definitely support you with that inside the membership. You know, and we're there to answer any questions and provide you with live help. We don't have a customer service bot in there. It is just Noel and I. So yeah, you're not going to get some unhinged answer from Ron Burgundy. Yeah. And really it's just an old custom GPT.
Noel (25:43)
Or is it?
Noel (25:57)
Yeah, absolutely. Yeah, thankfully I don't talk like Ron either, so yeah.
Katie (26:04)
I think that's a blessing for everyone.
Noel (26:08)
Very true.
Katie (26:09)
Yeah. So we will put the link below if you want to come and join us or if you've got any questions about AI Business Club, then just reach out to either of us.
Thanks so much for listening to this week's podcast. If you liked it, please leave a review or a thumbs up in the relevant place and we will see you for another episode very soon.
