The Art of Prompting ChatGPT: Craft Prompts Like a Pro

learned a lot more . Big shout out , by the way , to the biggest source of where I've learned so much about AI and tools and so forth is my friend and brilliant , brilliant mind on AI , rachel Woods from the AI exchange . She's also a former guest here on the podcast .

Also a little foreshadowing for you Got a big announcement coming your way in the not so distant future with regard to Rachel and the AI exchange , so I'm just going to leave it at that . So so many of us use tools like chat , gpt or Claude or Bard and they're great to help you come up with ideas right or generate content for you .

But one thing and there's a lot of things I recommend starting to think about when it comes to AI is using these tools and prompts to actually accomplish tasks in your business beyond just coming up with ideas or generating content for you .

But today we're going to keep it basic and talk about prompting to help you come up with ideas or generate something for you in the business .

In the Wednesday episode coming up here , I'm going to expand on this conversation here and talk about workflows and automations in your business that you can do with AI and other other tech tools , which I think is a really fun stuff , right ? So let's get just the basis out of the way , right ? Like what's a prompt ? Like what's a prompt ?

It's basically the input that you put into chat , gpt or another LLM , which is large language model , where you tell it what you want it to do or ask it a question or whatever . Like , that's prompting , and it's really an art .

And this is where I see a lot of people miss out on big opportunities for great results when they're prompting , because you're obviously trying to get the most accurate and relevant responses to whatever it is that you're wanting , right ?

Well , a lot of people will put together a prompt , put it in there into , you know , say , chat , gpt , and not be happy with the results , and they're like , oh , this doesn't work , right ?

Well , it's a skill , just like anything else , to learn how to create prompts that effectively communicate what you're trying to accomplish , effectively communicate to chat , gpt or Claude or what have you , and so that's what I want to help you here with today .

I want to kind of give you a structure when you prompt that can help you get the best results possible .

Now there's a couple of challenges that , among many other , that people face when , when prompting chat , gpt and you know , like I said , one of them is just finding the right prompt , it takes a lot of trial and error to learn how to craft that prompt in a way that you know gets you something that's helpful , that can get you something that you can actually

use . And , like I mentioned , a lot of people don't want to put in the time to learn sort of the nuances of prompting . And if you're going to get a benefit , if you're going to benefit from AI and these tools going forward , you have to . It is a prerequisite that you learn how to prompt .

I say that because right now , in in the not so distant future , you need to know how to do that Now . In the future , I kind of think that a lot of the tools are going to exist where you don't even have to create a prompt in there . You just tell it what you want it to do and it'll do it .

All right , but right now we have to know how to prompt . Another challenge is avoiding bias . A lot of these models , can you know , perpetuate harmful societal biases If you're not prompting .

You know very carefully , and so I want to encourage you that when you are crafting prompts , you want to be mitigating any kind of bias , and Bias in that means that you have to be aware and care about these things , and I know that if you're listening to my show , you do . Okay , so that's another thing to be aware of .

And the other thing , too , that you hear so often is some of these tools will kick out an output to you that is just flat out wrong , you know , or just makes no sense , or factually incorrect or inaccurate , and that's really where it's on you to be able to fact check .

And Rachel from the exchange likes to use the term and I love this having a human in the loop , meaning all the automation is great and creating these workflows in the tech stack , if you will , but you still got to have an actual person in there checking things . And , as I mentioned in the last episode .

What has not changed is that this is not a copy paste and that's the end , all be all . No , it needs to be reviewed . It needs to be checked , not only for accuracy I mean first and foremost accuracy but also like is it how you would write ? Is it , does it reflect you ? Et cetera , et cetera .

So those are some of the challenges and also what a prompt is . Now I want to share with you the structure of a prompt that is designed to get you the best output possible for what you're trying to achieve .

Now , the first step is super basic , and you've heard you've all heard this , I'm sure but you want to tell ChatGBT or Clawd or whatever , what you want it to be , tell it who you want it to act as . So , for example , when you go in the ChatGBT act as a world-class online marketer who specializes in creating high converting sales funnels .

So could you go right to a task like right to asking a question to accomplish something without giving it that sort of lead in and how you want it you know who you want it to be or how you want it to act ? Of course you can , but you are going to get far better results when you tell it what you want it to be or who you want it to act as .

So start there . I want you to act as a world-class gardener . I want you to act as a world-class veterinarian . I want you to act as an expert conversion copywriter , whatever it is . So sort of set the tone right from the start

. The second step is to give it context , any context that is going to be helpful for ChatGBT to know you want to put here next right . So like background or your target audience or your voice , you want it to have the context . So again , you're setting it up for success when it gives you results .

I believe in the last episode there I shared with you how to have it get your tone of voice etc . So just to kind of recap real quick if you have a podcast , if you do videos you know YouTube videos if you do right blog posts , right emails , what have you ? You can tell ChatGBT to analyze .

So if you have a transcription , for example , or a series of emails , you can have ChatGBT analyze that text for your tone of voice and your speaking style etc .

And then now you can put it into ChatGBT and it's called custom instructions so you can put your voice in there or your target audience , if you are always asking ChatGBT questions about your specific business , for example , so you don't have to retype or re-copy and paste that in there each time .

So this is the context section Any context that's going to be helpful for ChatGBT or Claude or whatever to know . The next section remember , this is all within a prompt , right ? So the next piece of the prompt is the criteria , and this is really , I mean , all this is really important , but this is really really important .

Tell it everything that you want and don't want it to consider when giving you the output , for example , in this example here . So tell it where you want the funnel to end up , like buying an offer , for example , and to be specific with what the offer is . You know it's my whatever AI coaching tool .

That's the final offer , but you're telling it , giving it criteria . Maybe you want it to use a specific funnel framework . Tell it that here in the criteria section of the prompt , maybe you're writing sales copy and you want it to follow a specific you know copywriting framework .

Or maybe you're creating a business plan or you're trying to come up with new business ideas or whatever you and you're using , like the blue ocean framework or blue ocean concept . Tell it that you want it to use that , because , again , this is we're trying to lead you to getting the best responses possible .

And then , in the final step of your prompt here remember , this is all one prompt , we haven't even hit the button yet to get results yet . Next is the instructions . Now , this is where you simply tell it what you want to accomplish here . This might be . You might also include , like how many options of results ?

Like , I want you to give me three different funnel options ? Blah , blah , blah . I want you to generate 15 email subject lines . Blah , blah , blah . Right , I wouldn't put blah , blah , blah in your prompt . Tell it how you want the results . Maybe you want the results in a table and it'll give you results in a table .

Maybe you want it to then break the project down . So if you're creating a funnel project from scratch , maybe you want it to then break the project down into weekly and daily tasks to accomplish that project and put those tasks into a table . You can do all these things . You're telling it in this last step of the prompt what exactly you want it to do .

So I'll recap then real quick Tell it what you want it to be or how you want it to act as , then give it context , like your target audience , your voice , any kind of background information , give it criteria , tell it everything that you want and don't want it to consider when giving you a response .

So in this high converting sales funnel , for example , maybe you tell it the goal of this funnel is for the user to purchase X , or maybe you want it to use a specific framework . Put it in the criteria section . And then the last piece is instructions . You're telling it what exactly you want it to do .

Now , if you go through these steps , you do have the option . If you want , always you can then say , before even you know , clicking the button to put the prompt in motion , you can ask ChatGPT if it understands what you're wanting to accomplish , and then you can ask it .

You know , ask me questions for how I might be able to improve the prompt that I've just put in , and then it'll ask you some questions . You answer those questions and then , once it starts giving you the output obviously you're judging the results at that point and you can then ask it to ask you X number of questions .

Maybe it's giving you three different funnel options and you're like , ooh , I really like number two . Well , you can have it , ask you five questions about funnel , the second option with regard to X , so you keep giving it additional context to continue to refine the output that you're trying to get to right .

Because you might say , oh , funnel , number two option that it gave me , I like that , but it's about 80% . I think we can improve that . So have it , ask you more questions in order to improve that funnel and you can give it additional context and rerun that prompt Again .

The idea is to try to get you to a point , obviously , where you're really happy with what it has kicked out to you . Okay Now , what I've just described here . You can totally do this straight inside of chat , gpt or if you go to Claude or barter or what have you .

We can also get a bit more technical here and say we're going to use the open AI playground where you have a bit more control over how you are setting the prompt up , for example , so you can set things like mode , the model , whether it's chat , she be chat , she be T3.5 , et cetera , and also the temperature .

We've heard the term temperature , so lower is less random meaning like it's very , very specific , kind of like just spitting back out to you what you've put in , and then the higher the number between that range is more sort of random and more creative . And so , but for this episode , what we're talking about here , we're sticking with .

As if you're simply going into chat GPT or Claude or barter , what have you right ? If you're like Rick , I still have no idea what my prompt should be in order to accomplish this task that I want it to do . Well , tell a chat GPT to be a prompt engineer and ask it what the prompt should be to accomplish X , whatever it is that you want to accomplish .

Tell chat GPT to be a prompt engineer . So there's the first step in the prompt . Tell it how you want it to act or who you want it to be . So if you're completely lost on what should this prompt be , tell chat GPT to be a prompt engineer and ask it what the prompt should be in order to accomplish whatever it is that you're looking to do .

And then the last point on this I'll wrap up here highly , highly , highly encourage you to create a prompt library . This is another thing that you know Rachel has really talked about from the exchange is create that prompt library . So why do we do this ? Number one I would imagine that you're using prompts over and over and over .

So if you are using chat GPT to create SEO friendly articles , that you don't have to recreate the prompt every time Hopefully it's just a copy and paste of the prompt and then you just sort of update it based on what specifically you're looking for , and so a prompt library can offer that for you and you can create this in something as simple as Google Sheets .

I use Notion and we have a prompt library setup in Notion , which is awesome . And the other thing that creating a prompt library does is it allows your team to all access it . Right . So access to the prompt library , which they can add their own prompts to it that they're using .

Maybe they can go in and say , oh , I'm gonna use this prompt here for writing an SEO friendly article , and they go and they use that prompt , and then it didn't quite get the results that they wanted , so they refined the prompt a little bit .

Well , then they can go back into the prompt library and refine it , so the team is able to access the library , update it and then refine the prompts as they start using them . So create a prompt library , however you wanna do it right .

And the idea here and this is where we're gonna get into the next sort of phase of this meaning talking about , like , putting the prompts together to create workflows and automations .

We're gonna dive into that a lot more in depth on the next episode , next Wednesday , but creating a prompt library is the basis that is gonna create a lot more simplicity in your business so that you can just keep going back to that rather than having to reinvent prompts every time .

So , if you have any , I know I've thrown a lot at you here in a short amount of time . If you have any questions about any of this , just shoot me a message on Instagram at RickMulready and I'll try to answer that for you . Cool , hope this

was helpful for you .

As I said , this stuff's changing all the time , but this , specifically , the structure of a prompt is what's been most helpful for me , and I think this is really gonna help you improve your prompts so that you can be getting a lot better results , more online to what you're looking for , and hopefully I've given you some ideas about some of the things that you

maybe hadn't thought about , which you can absolutely accomplish inside of ChatGBT or any of these other LLMs , right ? All right , my friend . That's all for today . Thanks so much for listening . I'll see you in the next episode . We'll see you then . We'll see you then .