A (5:05)

It's been a long time.

B (5:06)

It's been too long. In the intro, I was talking about, remember the days of our Regis and Kathie Lee where we. We would do this a lot, but we would talk a lot about Facebook advertising. Do you remember that?

A (5:19)

Yep.

B (5:19)

Okay. And if you look at our businesses today, we're not talking about Facebook advertising. Very different and especially for you. So I have a lot of questions to ask you today, and hopefully my questions will be really relevant. To our listeners. Like, I've been doing this thing on the podcast where it's like, you just come over to my house. I know you're in San Diego, I'm in Nashville, but pretend you live in Nashville, too. And you just came over and you sat down with me, and I'm like, okay, Rick, I've got some AI questions, and some of my questions might sound stupid, and you can't judge me. It's that kind of conversation.

A (5:53)

And my question back to you would be, how much time do we have?

B (5:56)

Okay. Yeah, exactly. I mean, I can't even imagine how much is swirling in your head, because as I was going to mention, you recently made a really big shift. I mean, is it safe to say you went all in with AI in terms of your business?

A (6:11)

Yeah, it's actually been like a year and a half.

B (6:14)

Okay. That's longer than I thought, actually.

A (6:17)

Yeah, it was. I mean, I would say officially. Officially, I'd say a little bit over a year. But I sold my podcast in the fall of 2023, and that was sort of the informal beginning, you know, unofficial beginning, if you will, of the new direction. But I went all in on AI in terms of content. I was like, you know, I. I mean, you know, you've been podcasting forever. And I was like, I've been podcasting for 11 years, and, you know, like, I want to try something new. So I went all in on YouTube, and I started officially late February of 2024, so just under a year ago. And YouTube's been amazing. So all in on YouTube, I went back to a membership. I've never been a huge fan of memberships, but it doesn't make sense, really, when it comes to AI to do a course because it's going to be outdated in two weeks. So, yeah, I have the online membership now and the YouTube channel and the newsletter. So I put out a very beefy newsletter every Sunday morning all about AI.

B (7:28)

But our. Makes me nervous when you say beefy, because I'm already overwhelmed, but, like, in the best way. I was telling you this when we just popped on, but as you know, Michael Hyatt is my business coach, and he's very much embraced AI, and he sent me and my friend Bonnie Christine a text this morning. It's from Julia McCoy. So Julia McCoy, she's very much. Also has embraced AI fully, and she said, your anxiety about AI, it's not fear, it's recognition. She goes on to say, your DNA knows you're about to transcend every limit, every boundary, every cage. This is what liberation feels like. I mean, this woman is going for it. And Michael says she definitely needs to come on my podcast. But that whole thing about your anxiety, it's not fear, it's recognition. I needed to hear that. And I'm just jumping in right now because I was telling you before we started recording, I have a goal of 20, 25 by the end of this year. I want to be fully proficient in AI in the way that it will fully benefit my life, business and personal, but probably more business. And I want to be able to navigate with ease. I want to know what's going on. I want to be excited. Excited, but also cautious about what's going on, but informed and be able to implement. And so when I say implement, like, right now, my main focus is learning custom GPTs. I want to be very proficient in that area. And so I tell you all this because I've been doing it for a few months now, but I really, over the last few weeks, got very serious about it. And I was in a group text with two of my girlfriends, and they were so much farther along than me. They were talking about things I didn't know. They were creating their custom GPTs with so much more ease than me. They knew how to, like, optimize them. And, Rick, I freaked out a little in my head instantly. I'm like, I'm so behind.

A (9:33)

But what were they saying? Do you like. No terminology or some. I'm. I'm just really curious, like, why you feel so behind.

B (9:41)

Yes, because we were sharing our custom GPTs with each other, and maybe we should back up and talk a little bit about some of these terms just in case my listeners want to get up to speed and aren't yet, but I was creating these, and we'll talk about what that means in a second. And they just knew what information to put in to get better output. They knew different formulas and things to ask it and ways to set it up so they can get better. The most out of it, where I'm like, oh, my gosh, how do you know how to do that? Or how did you know to talk to it like that? And it just, like, instantly made me feel inferior. And I just knew, okay, this is something new. I'm learning something new. I think it's normal. But what do you think about all that?

A (10:25)

Notice some of the things that you just said that I like to say that it's the whole. I mean, you've heard this for years and years now. Like, forget the saying. Exactly. But it's like, the power of the answer is based on like, how good a question that you ask or something like that. This is AI, because if you don't know how to ask the question of what you want to get done, it's kind of like crap in, crap out. It's not going to give you the output that you're looking for. And it sounds so easy, but that's. It's actually a really hard skill to be able to clearly articulate what it is that you want AI to do for you. Because like, hey, chatgpt, write an email for me. Okay? It's like, well, all right, it can do that, but that's not a very helpful statement. You have to ask a much more in depth question to get the kind of output that you're looking for. Does that make sense?

B (11:24)

It does make sense. And would you agree that not only ask a more in depth specific question, but also give really good context or background? And that's where I noticed yesterday when I was sitting down, I was creating. Let's. Okay, don't let me forget that thought I was creating something. I want to come back to it. Can you explain to people what is a custom GPT just in case?

A (11:48)

So I'm actually interested for you, jumping right to custom GPTs. Like, my question is, have you sort of quote unquote mastered prompting? Meaning like simply asking ChatGPT or Claude to do something?

B (12:04)

Okay, when you say master, probably not.

A (12:08)

Are you familiar with how to structure a prompt?

B (12:11)

Well, I want to say yes, but maybe not. So can we start there?

A (12:14)

Yeah.

B (12:14)

I mean, okay, I know we're a little bit all over the place. You know, I like to have structure, Rick, but I can't. My brain is like all over the place with this. So here's what we're going to do. One, we're going to talk about ChatGPT and Claude and what are the differences? Because that keeps coming up a lot in conversations. Two, let's talk about what it means to master prompts and really be able to optimize that. And then three, we'll get into chat GPTs and that's when I'm going to talk to you about how I've learned that. The description and the context, how important it is for me to understand what the heck's going on. So those are the things. Don't let me forget all those things.

A (12:54)

So where should we start when it comes to prompting? You remember when we used to talk about ads and someone would say, should I try this or will this work?

B (13:05)

Yes.

A (13:06)

What was our answer? It was like, well, yes, you should try it and test it because it all works. It's kind of the same thing with prompting. I did a video that got a lot of hate comments this past year or last year on my YouTube channel. I said, prompt engineering is dead. And the whole point of it was, is that these AI models. So a model is basically is GPT, is quad, is Google's Gemini. These are all AI models. They're getting so smart that if you are able to clearly ask the question that you want it to do, it will help you along the way. And if you don't know what to ask, you can ask it to help you. Meaning, like, hey, I want to accomplish whatever. Like, I need to, I don't know, create a project plan for my social media for the second quarter of this year. I don't know how to ask you for that. So can you please tell me how I should formulate this question to you? Meaning, like, you're asking ChatGPT, and people don't think about it like that. They just think like, ooh, I don't know how to ask that question, so I can't use it. And it's like, well, okay, got it. But don't forget, it's kind of like that. I always go back to that cheesy line in Jerry Maguire where he's like, help me help you. Yeah, it's the same thing for AI. And it's because it's like, if you don't know what to ask, you can ask, ask it and let it know. Like, hey, I don't know how to formulate this question. I don't know how to prompt you. Can you write me a prompt to accomplish X and it will do it for you. So my point is, is, like, it is very helpful, yes, to know sort of the basis of how to formulate a prompt. But if you don't know that, it's not the end of the world. You can just have a conversation with ChatGPT or Claude and start asking it questions. And then you just have a conversation to get to the point where it's giving you the output that you want. Meaning, like, in that hypothetical example, it could give you a, you know, a social media content calendar for Q2.

B (15:20)

Bonnie, Christine is one of my girlfriends that has been really diving in. And you and I both know her and love her, and, well, I love her. Do you love her?

A (15:29)

I don't know her that well, but I'm sure I do.

B (15:31)

You would love her if you knew her well. And I noticed it's really fun to get into text groups with other people that are learning as well because you ask great questions of each other and you learn things. And it was me and Jenna and Bonnie Christine in this group text. We were talking about it and she showed us some examples of what she was doing. And it was fascinating how casually she talked to her model in terms of the chat. Like, I'm still, you know, you know me super well. You know, I like to be kind of buttoned up, professional. And she's like talking to this person like it's a best friend that she's had for 10 years. And it was so cool to see.

A (16:10)

That kind of conversation and that really comes through especially. So, for example, if you pay for chatgpt on the. Well, it's actually on desktop now too, but on your phone you can. It's called the advanced voice mode, which basically just means like you're, you're Voice Texting, essentially. ChatGPT.

B (16:31)

Okay, I need this because I'm a master voice texter.

A (16:34)

Yeah. So you can have a conversation literally with ChatGPT without getting into the formalities of a prompt. Like before the holiday breaks here, my daughter's school is like two blocks away. I was walking home one day and, you know, it's like a six minute walk. And by the time I got home, I had fleshed out a new offer idea with ChatGPT by simply talking to it. I didn't prompt it, I just told it what I wanted to do. You can say, what answers, what information do you need in order to help me with this question? And it will tell you so to.

B (17:13)

Be very specific, chatgpt, it's on your phone. Is it a different model than the one I have on my desktop?

A (17:21)

No.

B (17:22)

Okay. So basically I'm just bringing it up on my phone. I already pay like $25 a month or something like that.

A (17:27)

20 bucks? Yep.

B (17:28)

20 bucks? Yeah. Okay, so I have it, then you.

A (17:31)

Do, and I'll pull it up on my phone while we're chatting. So in the. If you put up on your phone or anybody listening pulls up on your phone, you'll see the chat box and then on the right hand side next to the little microphone, there's like the little audio circle. That's advanced voice mode. You just press that and like the screen will turn to a large version of that circle and it will start, you can start talking to it kind.

B (17:57)

Of sounds like voxer. Kind of feels like voxer where you.

A (17:59)

Have a hundred percent, except it's talking back to you. I Don't mean like, you know, sassy, but it's talking back to you.

B (18:07)

I hope mine gets a little sassy.

A (18:08)

Here's a hint. When you do that, tell it not to interrupt you.

B (18:14)

Oh, see, this is the stuff I love to know about.

A (18:16)

Tell it not to interrupt you. Because if you like, pause for like a second to gather your thoughts, it'll come back in, it'll start to talk.

B (18:26)

Be like, whoa, whoa, whoa, I wasn't ready.

A (18:28)

Yeah, you will have to remind it. Usually I have to remind it even after I tell it not to interrupt you say, remember, don't interrupt me.

B (18:35)

So wild to me.

A (18:37)

And then keep talking. But going back to the prompting and just being conversational with it, it is helpful if you were on your computer to give it a role. Many people, I'm sure, have heard that. You've probably heard that. Give it a role. Tell it what you want, how you want it to act.

B (18:56)

You are a. Whatever, you're an expert in XYZ. Okay. So whether you're doing custom GPTs, which we'll talk about, because I do roles in that. But even just a prompt, you're saying, like, right when you get started, tell it what it is.

A (19:09)

Yep. And then you want to give it the task. Be very clear on the task. And you mentioned earlier, like, context. So context is super, super important because let's just say, for example, you want to write weekly newsletters and you want it to write it for you. AI loves examples, by the way. So any kind of examples, any kind of past work. Maybe there's a newsletter that you follow out there that you're like, I love how they write this. You can copy and paste that and use it as an example. It's not going to copy the word for word, but you can say, I like the format of this newsletter.

B (19:51)

Okay, I was going to ask you that next about these prompts. So you're saying feed your model and when you do a prompt, if you can show any examples, someone else is telling me they took literally a screenshot of, let's say, an opt in page they loved and then took the screenshot and just uploaded it into the prompt that they were doing. That's wild. It's not like, like AI can actually read from the image.

A (20:19)

Exactly. It's called multimodal, where it goes beyond text. Like being able to recognize beyond text. So it can understand images, it can understand audio now. And very quickly here it's getting to understand video. So that's what they mean by. When you hear the term multimodal that's all that that means.

B (20:40)

And I could say, let's say I was doing something for social media and I could say, write me a carousel IG post about XYZ and model it after this one. And I could put a link in there to a specific social media post.

A (20:55)

It's hit or miss. Sometimes when you put a URL in there, I would absolutely test it to be sure that it's going to work. I would just take a screenshot of it. Another hint, by the way, and I'm sure you've heard of it. It's the Go Full Page Chrome extension that will take like a running screenshot if it's like a long sales page.

B (21:17)

Oh, I love that extension. Yes.

A (21:20)

Yeah. So you could use that to grab like an entire sales page and then use that as an attachment.

B (21:26)

So it's like a Chrome extension. And it. I didn't remember this. It's called Go Full Page.

A (21:30)

Yep.

B (21:31)

Great. Okay, I have it, but I didn't know it was called that. So. Yeah, that's a good one as well. I love that. But go back to context.

A (21:39)

Yeah, so context is just basically like, what kind of background information or examples will help the AI give you the kind of output that you're looking for, that kind of information. So for example, your target audience. Great context. Because now the AI is going to be giving you. When I say output, it's just basically like the answer it's going to create for you based on your question. It will tailor it to your target audience, information about your business, like, what do you do in your business? That is a great piece of context, your tone of voice. So creating what I call a writing style paragraph is basically taking like writing sample that you have and getting the AI to essentially, I'll super simplify it, look at what your writing sample is and then create a paragraph essentially describing how you write.

B (22:35)

Oh, I love that. Okay, so I take maybe my best email or a few of them and then ask AI.

A (22:42)

Now I was going to say for you, you could. You could input your book.

B (22:44)

Okay, okay. Take my book. Oh, I get it. And say, give me a paragraph of my writing style.

A (22:49)

Yeah, I would say I want you to analyze the attached writing sample. And like in your case, you could say the attached book or anybody could just say, give it like five emails or blog posts or podcast transcripts. If you talk like you write, a lot of people write differently than they talk. And so I would say if you're going to be using this for writing, make sure that you talk like you.

B (23:14)

Write, okay, that makes sense. And then just to be really specific, it gives you a paragraph and you look at it and you're like, oh, this is really spot on. This is totally me. Maybe you tweak it a little. Now you take that. And I like to be old school in that sense. So let's say I put it into a PDF and so next time I prompt it, then would I upload that PDF into that prompt so it has it again? Or does it already have it so.

A (23:40)

It doesn't already have it? We'll keep it really simple because I can get technical here, but I'll keep it simple. Is the easiest way to do it is just keep it like on wherever a Google Doc or Apple Notes or whatever, where it's easy to copy and paste into your prompt.

B (23:57)

Okay. And say, here's my writing style.

A (24:00)

Exactly. Now, if you're going to be using ChatGPT to write in ChatGPT, there's the instructions in the upper right hand corner. You got your little icon of your face and your account. So if you go to it and you say, customize ChatGPT.

B (24:17)

Yep, I see it.

A (24:18)

It says custom instructions and then it says, how would you like ChatGPT to respond? You could put your voice paragraph right there. So now for every response that it's giving you, it's going to be writing in your voice, if you will.

B (24:34)

Okay, that's cool. Okay, so for everyone, just again, you go up to where your picture is in the upper right corner, click on that, and you'll see something that says customize chatgpt. I did not know anything about this, so this is why I love talking to you. That's a good one. But I love that you kind of pulled me back because obviously I want you to talk about custom GPTs, because I'm obsessed. However, you're like, whoa, whoa, whoa. Let's just make sure that we're mastering the prompts, which I definitely have not. And I could see why if I spent more time mastering how I'm using prompts, how much better I would be and probably wouldn't be struggling as much with the custom GPTs.

A (25:12)

Exactly. And here's a trick. Now I won't put you on the spot and quiz you.

B (25:16)

Okay, please don't. I'm not ready.

A (25:19)

But remember earlier I said if you don't know how to prompt AI again, I say AI again. ChatGPT, Claude, whatever it is that you're using, if you don't know how to prompt it, you can always ask it. You can do the exact same thing for creating a GPT. So the difference between just prompting and talking with ChatGPT or Claude versus creating a GPT is a GPT is going to be something that is repetitive, that you do all the time, whether it's daily, weekly, monthly, whatever. It is something that you're going to be using consistently over and over again.

B (25:59)

Okay, so let me give a really easy, great example. I've been working on an epic one for my podcast. So I have created two different GPTs. One is when I bring on a guest and one is when I do a solo show. So essentially, just so everyone understands, and I know a lot of people do, but there's newbies around AI Essentially, I'm creating this custom prompt for myself that it will literally I go into it and it will say, okay, let's create content for your solo podcast. And then I've programmed it to ask me specific questions. What's the topic, what's the air date, you know, whatever, different things. And then I've programmed it to give me a two page, essentially outline of different things I want. And boom, boom, boom, it spits it out. And so I could use it over and over again. I mean, I very much simplified it though.

A (26:57)

Perfect use case, because you could do that in ChatGPT, just going in there and chatting with it, but you'd have to input that information every single time you do it, which we don't want to do. The GPT allows you to just exactly what you just said, like go in, give it any kind of like, whatever the topic of the episode or what have you, and then it does its thing.

B (27:20)

And here's what I love. This is why I'm obsessed with these custom GPTs. So the cool kids just say GPT. They don't say custom GPT exactly. Okay, I want to be cool. Okay, so, so these GPTs, Rick, what I love is that. So for those who are a little bit new about this, I literally tell it, okay, first I want you to write, I want you to put the topic of the podcast. Then I want a summary of what the podcast is going to be about. And then under that I want 10 quotes that are related to this topic that you found online from these people or like, that's how specific. And so the outline always, it's a template now that I get every time I do a solo show. What I also love about GPTs is the knowledge base. So there's a section that's called knowledge where I can upload my summary of how I write. I Could upload my book, I could upload other books, I could upload every transcript I've ever done on a podcast. So it's now it's always there and you're not constantly feeding it because you've already done it.

A (28:18)

Exactly. And here's two kind of, I don't want to say tricks. It's just like helpful things to do because the knowledge base is really, like you just said, the key to GPTs. It really is. So the other thing that you can do is when you get a good output. So like when you get your two pages after you put in the necessary information, when you get it, you're like, that is spot on. I love that. Well, you could upload that to the knowledge base because now it has examples of the kind of output that you're looking for.

B (28:54)

I didn't know that. Okay, I love that. So once it gives me a really good one, that's going to be my example that I'll upload to the knowledge base.

A (29:01)

Okay, exactly. So that's number one. Like, whenever you're working with a knowledge base and you have great output, use that. Like put it back into knowledge base and tell it that this is a great example of what you're looking for. Again, AI loves examples. The other thing, and I'll send this to you afterwards, it's kind of creating a table of contents, like a, kind of like a one pager for what's in your knowledge base.

B (29:28)

Oh, tell me more. And whatever you send me, I'll put in the show notes.

A (29:32)

Yeah, so you're giving the AI sort of like a roadmap about what's in the knowledge base and then how to use it. So for example, you'd be like the name of the file that, like the example that you just. All right, this is a great example of what I'm looking for. You upload it into the knowledge base. You give it a name. Well, in this quote, table of contents for the knowledge base, you would say, like this file called whatever you named it is an example of the output I'm looking for.

B (30:01)

Rick, I love you. I didn't know these kind of things, so this is so cool. So you, you train it even on the. All the knowledge you gave it. You're kind of giving context around why you gave it that knowledge.

A (30:12)

Yeah, exactly. You're organizing it and you're kind of what you're. It's sort of like a. I mean, it's a kind of a table of contents, kind of a. An index, if you will. You're helping it navigate what information that you're giving it.

B (30:25)

Okay, so that's, that's a whole other level I have not done. So I will be doing that. And so what are you going to send me that I can put in the show notes? Like, what. That would look like an example of what that could look like?

A (30:35)

Yeah, exactly. It's, I think it's just a Google Doc now. If you're like, wait, I don't know how to create that, what will we do? We will say, here are the documents that I'm putting in the knowledge base. Here's an example of what I'm looking for in this table of contents or index. If you will create one of these for me based on all this information.

B (30:55)

That little nuance that you just said is where I'm not comfortable yet. Like, I'm not doing that. I get stuck and. Okay, so let's talk about custom GPTs. And you're putting in a lot of information in the. What's the box? Is it called Instructions?

A (31:11)

Yes. Yeah.

B (31:12)

You'll see like a field, a box that says Instructions. And that's where you put a lot of your background, your context, your actions, your input, your output, that kind of thing. And I'm putting it together and what you're telling me right now is sometimes I get stuck. Like, ah, how do I explain what this podcast is really about? I could go back in the chat, say I'm struggling with this, I need help to explain xyz. I could copy and paste that and put it in there. That's where I'm not using it enough.

A (31:41)

Yeah, I mean, going back to ads, people would put together an ad, they run it for four days and what would they say? Oh, this isn't working, I'm, I'm not getting any leads, or my leads are $50 and this doesn't work and they give up. It's the exact same concept. You have to like, keep testing the prompt or the instructions to get you the results that you're looking for. Just because it doesn't, maybe doesn't work the first time, doesn't mean that AI doesn't work. It's like you're not giving it the right information that it needs to give you what you're looking for.

B (32:15)

It's almost like practice makes perfect, meaning you've got to just keep doing it and doing it till you get comfortable.

A (32:21)

When you're creating a GPT, by the way, to create a GPT for everyone listening, go in the ChatGPT, click on Explore. Now you do have to have a paid account for this. Click on Explore GPTs in the left hand column. In the top right, go to the plus Create button and then you're going to see a two options. Create and Configure. What Amy and I were just talking about is the configure option. It's a little bit more of the advanced, but the Create option, it says on the screen, it says, hi, I'll help you build a new GPT. You can say something like, make a creative who helps generate visuals for new products or whatever. And then it says, what would you like to make?

B (33:03)

Oh, I didn't even know that tab. I've just been in the configure the whole time.

A (33:07)

This is where people are like, well, I don't know how to talk about what I want to make. And this is where it comes back to being able to articulate what it is that you want to create. And this is where you can tell it. I want to make a. What do you call your. The GPT that you just described for the podcast?

B (33:29)

I call it the one with the solo. It's called Podcast Solo Content Assistant.

A (33:35)

Okay, cool. So you could just be like, I want to create an assistant that helps me put together notes for and prepare for my podcast episodes. You could just type that in there and then it will ask you the questions that it needs answers to in order to create that.

B (33:53)

I probably should have started there, but now I'm working with Michael, who's very techy, and so he taught me how to configure, which I love that I know how to do it now. But if I didn't have Michael to teach me, starting with Create would have been the best idea.

A (34:07)

It is the easiest because you don't have to look at the fields within the configure option. Be like, I have no idea what, what box do I uncheck? Or whatever. You can just click on the Create option and just. This is where you're just having a conversation. You don't need to prompt it. You're just telling it what you want and it will ask you questions when it's ready for you to test. Know that as you are telling it what you want and it's going back and forth with you, it's writing the instructions for you. So when you go over into the configure option.

B (34:39)

No way it shows up there.

A (34:41)

Yeah, it shows up there. It's writing the instructions for you.

B (34:45)

Okay, so this is so great for all of you who haven't created custom GPT yet that do it that way because then you'll see what it looks like to configure your own.

A (34:55)

Exactly. And then if you're like, wait, this isn't giving me what I want, just go back to the Create tab, if you will, and just tell it to say, hey, this isn't whatever. Fill in the blank. And it will rewrite the instructions by itself until it gets to a point where you're like, oh, I'm really happy with this.

B (35:16)

I know this sounds dramatic, but that kind of gets me so excited. It's like so cool to learn something new. It's been a long time since I've started from scratch learning something brand new, so it really is cool. Now I'm going to ask a super techy question that I know I can Google, but since I've got you. And so those of you who have no idea what I'm talking about, once you do your first GPT, you will see this and you will be so glad I asked Rick. So, Rick, when you go to the bottom of Configure, there's this place that says capabilities and there's four things you can click without getting super techie. What in the world? What am I clicking?

A (35:58)

It's a basic question, but a super important question. So listing out. I'm just reading this here. So the four things that it lists out right now until OpenAI changes this, it says web search, Canvas, Dall E image generation, and then code interpreter and data analysis. So what the heck does all that mean?

B (36:19)

Yeah, and this is where my anxiety comes in. It feels like, okay, I'm really bad at math and it feels like, like code or math or stuff I just don't know about. And then I'm like, oh my gosh, this is over my head.

A (36:33)

Yeah, and when it says code interpreter and data analysis, there's a little question mark there that you can hover over. And it just says, allow this GPT to run code when enabled. This GPT can analyze data, work with files that you've uploaded, do math, and more. So basically, if you are uploading files into the knowledge base, you're going to want to make sure that's checked because it needs to be able to analyze that information. So what I like to do is, number one, turn that on. If I'm not working with images or I don't need it to create any kind of images or anything like that, I will turn off the Dall E image generator. Canvas is. They just added this. I mean, we're recording this the first week of January. So middle ish of December 2024, they just added the canvas to it. I would leave it checked. It's fine.

B (37:25)

What does that mean? What does canvas mean?

A (37:27)

So canvas is when you're working. Have you worked in ChatGPT yet? Like the regular chat. And then another box pops up to the right hand side. So it kind of like gives you the results as you're typing in.

B (37:41)

No, I haven't.

A (37:41)

It almost like, it almost like splits the screen.

B (37:44)

Okay.

A (37:45)

That split screen option is Canvas. It kind of allows you to see sort of results in real time. It's mainly for if you're writing or coding. But I would leave it. It's no problem to leave it checked now. Web search is a different story. Out of all of these, the two most important are code interpreter and data analysis. If you are uploading information to the knowledge base, which I think you should if you're using a GPT.

B (38:15)

I just clicked it. Yeah.

A (38:16)

The other option here of web search, one of the biggest complaints that I hear from my members and from people is I've set this GPT up, I've uploaded files into the knowledge base. But it's not using the knowledge base to answer the question that I'm giving it, it's using like ChatGPT. The whole reason that people are creating a GPT mostly is to access just the knowledge files. So in that case, I do two things. Number one, I turn off web search. But then I also, in the instructions, I tell it when answering, only reference the knowledge base files. Do not use the Internet. Whether you're in the configure option, I would put that into the instructions. Or you can just go back to the Create tab and tell it, say, and I was creating one right before the holidays and I probably had to go back and forth five times with it until I got it to a point where it was not referencing anything outside of the knowledge files.

B (39:25)

Okay, that is huge. And, and those of you who are like, what are you talking about? I promise once you get in here, you'll remember this conversation and be like, oh, I'm so glad they talked about that, because I genuinely did not know. I have one that's called Leadership Assistant and I uploaded, it's a GPT and I uploaded all of my most favorite leadership books and I created it for my leadership team to ask questions when they get stuck with employee things, conversations they need to have hard conversations, hiring, firing, everything in between. But I'd rather it not go on the web. I'd rather really use these Books that we swear by. And so I just went in there and I just unchecked web search. But I'll also go into configure and say, what was the term you used? Do not use web. Or what did you say?

A (40:14)

It's again, the language, you're going to have to kind of play with it. But I would say when answering, only use the files that I've uploaded into the knowledge base. Do not use the Internet.

B (40:28)

Got it. That's so interesting. And I know some people are like, why wouldn't you want it to use the Internet? I can see pros and cons of both. But also, you know, there are some scary things on the web that I don't know what it's pulling from. So I'd rather pull from some sources that I've already used. Like let me give you an example. One of the books is Radical Candor. And so that book we swear by in my company, I'd rather pull from Radical Candor about a hard conversation than some random website I know nothing about.

A (40:56)

That's a great example of how can I create something as a resource that either I or my team can go back to later. This is where it's like, get creative as possible. Creative as you, as you want. So if you have customer service FAQs that you get asked, create a GPT out of it. So somebody on your team, if you're training them, for example, they can just go in here, they don't know the answer to a question, they can just put it into the GPT and they can, they'll get the answer for it or put in your, you know, your employee handbook or anything like that where people can be like, oh, what's the, I don't know, what's the paid time off policy or something? Go into the GPT answers that, okay.

B (41:45)

On those lines I've got to share one that I think I'm so most excited about. So, you know, we launch Digital Course Academy every year and we bring on support team members that are just seasonal, just for Digital Course Academy because we get so many questions over a two week period. And so we train them and we get with them like two months in advance and we start working with them. But there's a lot of information in Digital Course Academy and there's a million questions people could ask. But now I get to create a GPT where I literally put the entire transcript of the entire program in there. I put every PDF that is offered in the program, I put every answer to every faq we've ever collected. Now this seasonal support person is going to kill it. And that's the kind of stuff I'm excited about.

A (42:34)

Exactly. The whole book. 10x is easier than 2x.

B (42:37)

Oh, one of my favorites.

A (42:38)

So I created the GPT based on that. So, like, now I can use it as a brainstorming partner, if you will, that is trained specifically on that, that ideology, that framework, that information.

B (42:53)

Okay, that's a good one.

A (42:55)

Another trick on the GPTs, by the way.

B (42:57)

Oh, give me all the tricks. Okay, I'm ready.

A (42:59)

All right, so at the very bottom, when you are creating your GPT.

B (43:03)

Yes.

A (43:04)

In the configure option, if you have uploaded files. And chatgpt is very sneaky with this. If you've uploaded files, make sure you scroll all the way to the bottom of the configure screen under Actions Create New Action. Do you see like an advanced settings?

B (43:24)

I do. It says additional settings on mine or additional settings.

A (43:28)

That's what it. Yeah. So click that. Is it checked?

B (43:31)

Yes.

A (43:32)

Uncheck that.

B (43:34)

Why? Okay, so just so everyone knows, it says use conversation data in your GPT to improve our models. What does that mean?

A (43:44)

So it's basically saying any information that you put into this GPT and that you're, you know, when you're using it, all that information is being used to train ChatGPT's model.

B (43:55)

Okay, so I thought that was not going to happen because in settings there's also a place that is like, don't share everything I do to improve the model.

A (44:06)

Yep.

B (44:06)

Let's just make sure. Let's just. Let's do both.

A (44:08)

Exactly.

B (44:09)

Okay, exactly. So everyone just knows what I'm talking about because I know this is a hard conversation when you're driving and you're not. Not looking at it. So if you get a chance to come back to this conversation and be in front of your computer while we're talking about this stuff and opening it up, it will be more valuable. And sometimes I like to listen to things twice. Anyway, so if you go to your little image, your photo in the right corner and you go to Settings, you're going to look for data controls and you'll see it says something that says, improve the model for everyone, that that's turned off. And also out of respect to the books that you upload as PDFs, you've bought those books on Kindle. And so out of respect to those authors, you don't want to be sharing everywhere, obviously. So I think that also is in integrity.

A (44:57)

Yep.

B (44:58)

Okay. I mean, are you impressed about How I can have this conversation with you.

A (45:02)

Rick, I'm very impressed. That was very good.

B (45:06)

I will tell you this. I'm kind of teasing, kind of not, but weeks ago, I couldn't have said half of this to you or even asked these types of questions of you. And so for those that are listening, I just dug in. I just said, okay, this is so awkward to me. I feel very behind. I want to figure this out. Let's go. And if you get a few friends that are doing it with you, I promise you that is so valuable. But also just to give Rick a little plug. Rick, you have a membership that has been very valuable to me as well. I'm not done with my questions. I've got a few more. But can you tell people real quick about your membership? Because that's a great place to start, too.

A (45:43)

Yeah. I mean, we're all going through this together. And that's why I created this. Because AI is changing so fast. And like we said before, like a course doesn't make sense for AI because it's outdated in a matter of, you know, a couple weeks. That's why I created the membership. And you mentioned Michael Hyatts in there. I mean, he's a great resource. And it's for online businesses and who want to learn AI. And how do you use AI in your business? So that the whole idea is like, how do I streamline the business so that I am working less in the business while increasing profit? And I know that that sounds all fluffy and cliche, but you don't have to spend as much money on different things in the business when AI is doing it for you. That's what I want for people. So that's why I created it.

B (46:36)

Amen. Okay, so what's it called?

A (46:38)

It's called the AI Playbook Premium Community.

B (46:42)

Okay, and where do they go if they want to check out your membership?

A (46:45)

They can go to AB, porterfield.com forward slash. Rick.

B (46:50)

Perfect. Okay, so I highly recommend it. I'm in there and I love it. Okay, so can we go back to that one question I had about Chat GPT and Claude? Because I've been hearing a lot about this, and then I'm like, oh, my gosh, I can't learn another model. And I've been putting a lot of stock into this Chat GPT. What if I have to do both? Can we talk about the difference? And do I have. I say have to. But do I need to be using both?

A (47:16)

Yes, you do.

B (47:17)

No. The answer is no. You're breaking up. I Can't hear you anymore.

A (47:20)

You're breaking, you're breaking up. So, Claude, no one wants to hear this because for that exact response, like, I don't want to learn another model, right? Claude is a superior writer of content. So I use Claude to write any content, emails. If you're going to write a sales page, if you're going to write a landing page, if you're going to write social media copy, Claude is going to be your best friend. Claude also sounds, even without putting in your voice paragraph or teaching it your tone of voice or anything like that, it sounds funny to say this when you're talking about AI, but it sounds the most human, it sounds the most empathetic. When you're working with it, Claude will become your best friend.

B (48:12)

Really? I don't know. Let's talk in a few weeks and I'll tell you.

A (48:16)

I don't use ChatGPT for any kind of writing.

B (48:20)

Interesting.

A (48:21)

Okay, I use ChatGPT. It's great for brainstorming, it's great for research, analyzing information research, all of that stuff. I absolutely use it. But for writing content, even if it's like webinar slides or what have you, I'm going to use Claude.

B (48:39)

All right, so here's my question. Do I have to learn a whole other language like I just learned with configuring a GPT?

A (48:47)

So Claude calls it projects. So a cloud project is akin to a GPT.

B (48:55)

Okay, so a project would I. Let's say I want a new email onboarding sequence. So anyone who comes into my world, I want them to be part of a new email onboarding sequence, I go into Claude, do I set up a project for that?

A (49:10)

So let me, I'll ask you, is it repetitive?

B (49:13)

Like, no, I'm only going to do this once.

A (49:15)

Exactly. So I wouldn't create a project for it. I wouldn't create a GPT. So again, if you're like, I don't know how to prompt this, if nothing else, give it a role. So you could say you are a world class email funnel writer or email sequence writer with a deep knowledge of fill in the blank of whatever your niche is and then tell it like, I want you to do this, but I'm not sure what, like how to ask you or what questions or what answers that you need or what questions that you need answered. Please let me know that part.

B (49:53)

That again, like a nuance is so valuable. Like that might be the most valuable thing you've offered. And I know you know a lot more than that. You're like, thanks. Like, that's the one thing. But just because a lot of people, including me, I'm like, I don't even know how to put this together yet, because. Okay, so don't let me forget. I want to come back to something you said about a project or a GPT when you need it or not. I have an example I want to share with you, but I want to come back to putting together the information for a prompt. I've noticed AI has made me better at really understanding who my audience is, what my brand is, what my voice is, because I'm telling it. And I'm trying to be very thoughtful and intentional. And I've gotten stuck a few times, like, oh, my gosh, I should be able to rattle it off my tongue, what my podcast is all about. But something felt like it was off. I think that makes us better as marketers to really dig in at that level.

A (50:52)

Absolutely. It shines a light on what you don't understand as well as you thought you did.

B (50:57)

Yes. And getting. You'll get better.

A (51:01)

Exactly, exactly. And it's only going to improve things. So I hear from a ton of people are like, well, I don't really know. I don't really fully understand the pain points of my target audience. Well, use ChatGPT to help you with that, by the way. Very important distinction. Very important distinction. Claude is not connected to the Internet.

B (51:23)

Whoa, whoa, whoa.

A (51:24)

ChatGPT is connected to the Internet.

B (51:27)

That's incredibly important to know.

A (51:30)

Yes.

B (51:30)

So you need to feed Claude.

A (51:32)

Do you feed Claude knowledge in a project? Now, here's the other big difference. The amount of knowledge that you can put into a Claude project is much smaller than a GPT.

B (51:46)

So you want to be really intentional.

A (51:48)

Yes. Yeah. And I could get into other technical tricks there to like, sort of expand how much you can put in there.

B (51:56)

But I don't join his membership.

A (51:58)

Different conversation.

B (51:59)

Yes. He teaches that in his membership. There's ways to get around it so you can. You can put more in there. There's tricks. I do know that. Okay, so let's go back to when you need, let's say a GPT. And remember, everyone who's new to this, it's like something that you're going to keep coming back to again and again, such as my podcast show notes or my podcast guest prep versus just using a prompt. Okay. And we talked about that email onboarding sequence. I don't think I'm giving away too much. I'm going to use an example from Bonnie Christine, and I want to say her name, because this was her idea. She created a GPT that is basically called Bonnie Christine Marketing. And she fed it knowledge around, you know, her brand, her brand story, all these different things, really intelligent about her business. And then she also fed it her favorite marketing books. And so she used that GPT to then say, okay, I want to create a new email onboarding sequence. And it was pulling from all this knowledge. So in that case you could just use a prompt and say, hey, I want to create a new email onboarding sequence. But you also could have this GPT that will, in my opinion, make it even better.

A (53:15)

So notice the difference there. Is that the GPT, how she's created it?

B (53:22)

Yes.

A (53:23)

It's being set up to go back to constantly for different tasks.

B (53:31)

Yes, yes.

A (53:32)

Right. Now that can work a hundred percent. Best practice is to have single task, single GPT.

B (53:44)

Why?

A (53:45)

Because if you're creating a GPT and you're giving it instructions of like, you're going to write onboarding sequences, you're going to write sales pages, you're going to write ad copy, five different things. Well, those are five different skill sets. Writing ad copy, for example, is very different from writing a sales page. And writing a sales page is very different from writing an email onboarding sequence. So for that reason I would stick single task, single GPT.

B (54:14)

Interesting.

A (54:16)

However, however, here's the whole, like you gotta test it. What she's done here is she's created this massive base of knowledge, if you will, on the business with the ability to use all that knowledge to answer any question or have it do whatever she wants it to do based on all that information.

B (54:41)

And I think that's powerful.

A (54:43)

Now, early December, I did a full day coaching for somebody that I used to coach and they like to come back every end of the year to plan out the following year. So I coach, I coached them for nine hours that day.

B (54:57)

Oh, wow.

A (54:58)

So what I did was I took the nine hours of transcription of our calls. Cause I recorded it. I created the GPT around it.

B (55:09)

Oh.

A (55:09)

So I had nine hours of coaching transcription and I created the GPT for them so that it had all the stuff that we talked about and that I coached them on and the discussions that we had. So now they can go back and not only can they ask it, like what did Rick say about whatever it'll pull up in the GPT, but also anything that we talked about specifically, for example, an onboarding sequence, which is, we actually talked about that it can write an onboarding sequence based on all of that information that we talked about and information about that person's membership. So it's similar to what Bonnie's doing, but a little bit different in that all I did is I took just the, you know, nine hours of transcription, which was whatever number of pages it was, and made a GPT out of it.

B (56:11)

Okay, so the greatest part about this conversation is there's so many different things you can experiment with. But I think the biggest lesson and takeaway is you have to get started. Because, Rick, I had a series of questions to ask you and I did not get to any one of them. And I just looked at the clock and we're almost an hour. And I don't even do podcasts that are an hour usually. So, number one, I loved this conversation. Number two, I think you're gonna need to come back for the record, if you're cool with that. But I wanna actually kind of wrap up here because I really do think that I know I'm in it and I'm doing it and I'm new. And everything you shared in this episode was very applicable of getting started and really understanding how to use the models. And I really do believe that anyone that's already started that wants some new nuances, tips and tricks, they just got it, or anyone who's just kind of feeling fearful of getting started now, you know you're going to jump in and this is all going to start to make sense, like in a few weeks. I literally understood everything that Rick said, and I want to be clear. In December, Rick presented at Entrepreneur Experience for my digital course Academy graduates. I didn't understand a word he was saying weeks ago. And so that's how quickly you can learn this. And I want to encourage everyone who's listening, just sit down and get started and spend a few hours a week, if you can, educating yourself. And you're also going to fall in love with it and it's going to get exciting. And I know there's pros and cons of AI, that's a whole other conversation. But using it in a way to be more effective and productive, that's really where I'm focused right now. And just to be a better marketer. So with all of that, first of all, again, I think everybody should join your membership. And for the record, this isn't an affiliate code. I just wanted Rick to come on and I just wanted him to tell you all he has a membership. It's fantastic. Amyporterfield.com forward/rick, if you're just starting out or if you've been at it for a while. Do you agree, Rick? It's not just for beginners.

A (58:15)

No, not at all.

B (58:16)

Actually, some of it kind of scares me, so. But it's my aspire, so it's good to be a part of those conversations.

A (58:22)

That's the thing is what I was going to say is like just in talking with you when you were telling me that you felt behind, based on what I'm hearing from you, you're not behind.

B (58:31)

That was like nicest compliment ever. Thank you.

A (58:34)

And not to diminish what your friends are doing on your calls and stuff like that. That's basic stuff.

B (58:43)

Whoa, the girls are coming after you. You just called us basic. This you did, you did, Rick.

A (58:51)

But what I want to point out though, the value of that is the ideas that come up from those conversations, like, oh, I didn't know I could do that. Or that's a great idea. That's the thing that I think is most valuable in these conversations is like, oh, I never even thought about, whatever, uploading a Kindle book that I have. And I've always wanted to coach, but I couldn't afford a coach. But I love this book talking about coaching. Well, I can create a GPT based on that book and now I have a coach that I can access whenever I want.

B (59:29)

It's huge. You're right. Those conversations about what is possible.

A (59:32)

Yeah, exactly.

B (59:34)

And then give me a little hint. Don't get too nerdy here. Give me a little hint. Three quick fire things that are going to move me into being intermediate once I'm beyond the basics. Rapid fire. Ready, go.

A (59:50)

All right, so to move into intermediate, this sounds basic, but it will get you there. Look at your business. What are all the things that are being done within the business that are repetitive, that are happening over and over, whether you're doing it or the team is doing it, Setting up AI automations or systems that can take care of those things, that is what's going to free up time. And by the way, I should have said this in the very, very beginning, is like nothing that AI gives you in terms of output should be copy and paste. So it's not like, ooh, Claude wrote me this great onboarding sequence. Copy paste into Convertkit or whatever tool. No, no, no. You want to take that and then you put your personality into it, stories, whatever. You do any kind of editing and then you use that. It's not copy and not copy and paste. So go through the business, identify those things of what you are doing on a repetitive basis. That's where you start to look at it in terms of, oh, what can AI do to solve the problem? Because most people look at it from, ooh, chatgpt, cool. What can I do with it? And that's backwards. It's what problem am I trying to solve? And then figuring out the tool that is best suited to solve that. Because as we establish, they're like, oh, I want to write 90 days worth of weekly emails or I'm just whatever, or a nurture sequence that's 90 days long. Oh, I just found out that Claude is a better writer. And so I'm going to figure out Claude and how to use it to write 90 days worth of emails in 10 minutes. So we're looking at the problem first and then what kind of solution can we, can we create? I think getting really good at these GPTs that we've been talking about. Also learning Claude, that's going to move you more into the intermediate. And I would say the next thing you definitely want to do, and I know that you love to do voice texting, use advanced voice mode on ChatGPT. And that is as simple as, like, I'm out for a walk and I want to have a conversation and flush through some ideas. The cool thing about it when you're doing that is chatgpt is taking all of the. The transcript of everything that's being said and it will show up in your ChatGPT account so it's not lost. So I would say that's the next thing.

B (62:23)

Right. We'll go more intermediate the next time I bring Rick on. But this was one of my most favorite chats and I've never done an interview, Rick, where I didn't ask questions. One question from my outline. And so I actually am really proud of myself.

A (62:38)

I'm proud of you too, for going with it. Cause I know that's, that's not always easy. The other thing I'm going to do, and this isn't going to surprise you, that I created this, but right before the holidays, I created. Now follow me here. I created the GPT that will write GPT instructions for you.

B (62:56)

Oh, my God.

A (62:57)

Because this is the problem, like everything we've been talking about today, people have this problem.

B (63:01)

Yes, it's the big one.

A (63:03)

What kind of what to write in the instructions. So naturally I wrote it. I wrote instructions. I created the GPT that writes GPT instructions for you. All you have to do is tell it what you want it to do.

B (63:15)

Okay, so why are you teasing that people are going to want it?

A (63:18)

I'll send you the. I can share. I'll share the link with you.

B (63:20)

Oh, can we put it in the show notes?

A (63:22)

Totally. Yeah. Yeah.

B (63:24)

Okay. I'll be putting. That's huge, my friends, because I have spent hours and hours learning this that something like that would have been very valuable. So we are going to put that in the show notes as well. So amyporterfield.com749 that's where you go. It will have a link to Rick's membership, it will have a link to that index we talked about with knowledge, and it will have a link to a GPT that you can use that will basically prompt you on how to create your own GPT.

A (63:54)

My brain has been off for two weeks over the holidays here and I am so pumped up now after this conversation. I just want to go use some AI and let's go.

B (64:04)

Let's go play around with AI. Let's do it. Oh, my goodness. Thank you, my friend, for coming on. Definitely one of my favorite conversations. I really appreciate you.

A (64:13)

Thanks for having me on. This is fun. Appreciate it.

B (64:15)

Okay, so there you have it. I hope you love this episode. I really genuinely love chatting with Rick about all things AI. But listen, this episode is one of my longest ones I've ever done and so I won't talk too long. Now, what I do want to tell you is we've got a lot of good things in the show notes as you've heard on this episode. So go to amyporterfield.com forward/749 and I'll hook you up with all of those links. All right, my friend, I'll see you soon. Have a great week. Until then, take care.