The STEP Your Missing To Get More Consistency Out of AI

Dan Sanchez [00:00:05]:
Hello, and welcome back to another episode of the AI Driven Marketer where I, Dan Sanchez, friends call me Sanchez, am on a journey to master AI in 2024. Today, this is going to be part 2 of a course I'm just releasing via the podcast for free here on how to master the AI fundamentals. I'm 4 months into the journey, and I promise, like, I wish I woulda had these fundamentals early because they are the secret to success for everything else I'm finding out. In the first part, I covered how to build super prompts, and that's gonna be important. If you haven't listened to that episode or watched that video in the course yet, go back to find that one, and listen to it first because I'm picking up where I left off from that, that part. This is going to be part 2 where I'm gonna be covering the step method for how to build get better results with AI. And it kind of has to do with prompting, but it's slightly removed from just better prompting. So cover that in the last part.

Dan Sanchez [00:01:00]:
Come to this as part 2 on the step method for how to get way more out of your results with AI. So here we are. I'm gonna dive into this video now. And if you're listening, I'm gonna try my best to describe it as we go along here. But just know that I am actively flicking through on the video here. If you're on Apple Podcasts, you can follow along, or on YouTube, of course. If you're on Spotify, then you'll have to jump over to one of those 2 platforms or aidrivenmarketer.com where you can follow along with the video. But I'll do my best to describe it as we go.

Dan Sanchez [00:01:29]:
I'm here in OpenAI's Chat GPT 4. This is just a plain vanilla Chat GPT, no custom GPTs yet. And I'm looking at the last one we did where we built this huge super prompt loaded up with the podcast transcript in order to turn it into a much nicer post for LinkedIn. Right? A very common task for b to b marketers and marketers in general is turning some kind of content into another piece of content. And today, I wanna talk about how to use that step method in order to do this process even better. So let's take a look. This is actually a post that I posted on LinkedIn just a week ago, and it kinda it didn't do so well, which is too bad. Sometimes you post your best stuff to LinkedIn and just it just doesn't go so well.

Dan Sanchez [00:02:09]:
It's probably gone. It got 700 views. That's okay. That's just kinda how it goes on LinkedIn sometimes. People like the fluffy stuff. They don't like the deep stuff. So I'm I'm rehashing it here on the podcast for you guys because this is good. This is stuff that I'm using in AI more and more often.

Dan Sanchez [00:02:24]:
So let's cover what the step method actually uses. And this is a method that I've developed because I just found I was doing it and come up with I came up with a fun acronym so it's easier to remember. So let's talk about what the letters and steps stand for. For 1, we start with S, which stands for starter content. Then we move to T for template, E for example, and P for prompt. Now if you watch the last part of the first part of this course, we've already covered 2 of these things. Right? We've already covered prompt, which is what I'm for super prompt. If you come back to this last one, it's got this really long prompt at the beginning.

Dan Sanchez [00:03:01]:
Right? It's not just a short, hey, turn this into a LinkedIn post. It's it's very specific in what it's asking for, and And we broke that down in the last one, so we're not gonna recover it again here. We're also, in that one, put in what I call the starter content, because you can't you can ask AI to come up with content on the fly. Like, hey, write me a post about biology. And it'll be like, okay, it'll write you a post about biology. But if it's just pulling out of its own database or pulling it out of its head, then it's it's not gonna be that good. But if you're an expert in biology and give it something to start with, like I did with this podcast transcript, from a client's podcast that I manage. The podcast is closed mode, and the guest was Matt Brownlee, and they talked about discipline in sales.

Dan Sanchez [00:03:41]:
Very fun. Then you have its AI does great at repurposing it. And the step method brings you even closer because it adds in 2 other elements that I didn't talk about in the last one, and that is giving it a template to work from and an example of that exit template executed a plus. Right? So let's take a look at some what we can do with the template and example. Now I could have made a template and example to as an example for this video, but instead, I was lazy, and I just Google searched LinkedIn post templates. And this post popped up from Adrian Scheel. Adrian, I'm gonna have to, like, at mention you in this and make sure that you get your credit where credit's due. But look at look at Adrian Scheel, 5 simple templates for crafting the perfect LinkedIn post, because you had a bunch of good templates here.

Dan Sanchez [00:04:26]:
And I I copied and pasted 1 and used it. But I wanted to give you credit because this is where I got these templates, and I specifically picked your template 5 here. There are 3 things. He has 4 other ones, but I'm just gonna use this one as the example for the rest of this podcast and this lesson. So here, here's Adrian's version where he has a fantastic template. And I love this template because it's highly specific, yet uses like this almost code language to fill in the gaps. So it's like there are 3 things and then in brackets that will help you do that help that will help do achieve, there's like a typo here, help you achieve an awesome outcome. Right? Wanna know what they are? Here you go.

Dan Sanchez [00:05:09]:
In code brackets again, first thing, different code brackets. Reason why you should do first thing, code brackets. 2nd thing, code other code brackets. Reason why you should do second thing. Right? So it's very clear what to put where in this template if you know what those three things are. And AI, I found is very good at looking at this like Cody type language, a shorthand for insert thing here. This isn't actually code, but I think we've all seen code enough to know that, like, hey. This means insert something here.

Dan Sanchez [00:05:37]:
AI is really good at interpreting that and actually moving forward with that. So I love looking working with templates like that. When I build templates, I build them the same way. And underneath, we have a fully executed one that I believe he's written, probably posted himself. And I just went and copied both of these into my own doc. You could actually just take this whole doc and take it as a PDF and upload it to your GPT, but I'm gonna make my own because he's got multiple templates and examples in here. So I'm just gonna simplify it a little bit just to have one of each. And I made a new Google Doc called LinkedIn post template, and it's got 2 headers in it.

Dan Sanchez [00:06:11]:
One says LinkedIn post template, one says LinkedIn post example. Keep it simple for the AI to look at so that it doesn't get off track. And I just copied his template, and I put his example below. I only made a slight modification because I still want my template to say you can find my full interview at close mode dot media. Right? And I have that below in the example as well. So let's take a look at what this looks like when I give AI this template an example alongside my transcript and alongside my prompt. Now to get this into AI, you probably take a file like this, and down I'm gonna download it. And I already had already downloaded it as a PDF.

Dan Sanchez [00:06:50]:
And then I went over to chat gpt and copied and paste my same quote, and I reworked it a little bit. I also attached this PDF. You just drag and drop it right into the message area when you're crafting this prompt. And I changed the prompt a little bit because now I'm really loading it down. I mean, this is a beefy freaking prompt. It's got a PDF. It's got a whole transcript pasted in there. The prompt's freaking long.

Dan Sanchez [00:07:11]:
So to just to make sure that there's no room for error here, there's no confusion for the AI, I made it clearer and simpler by adding, like, 2 like, breaking it down into step by step instructions and giving it essentially 2 steps. And now it's crafted to say you are an expert sales coach and former VP of Sales Matt Brownlee. Take a deep breath and follow these step by step instructions. You'll notice this whole deep breath thing is very popular thing in AI prompting if you've ever explored this topic a lot, and I find it does work pretty well. And I use it anytime I'm having AI do step by step instructions or doing something that's like, there's a lot of elements to consider all at once, and I needed to think through it carefully. I tend to take do the take a deep breath and follow the do this. And then I broke it down to steps. It was essentially the same prompt, and I just broke it down into smaller pieces based in the first step is based on the transcript from the episode, find one solid idea in the episode, then move on to the next step.

Dan Sanchez [00:08:08]:
Step step 2, turn the idea into one actionable post for LinkedIn using the template and example in the attached PDF in quotes, LinkedIn post template dot PDF to guide your writing. Very important that you make it super clear, kind of like how you'd be writing to a genius in turn that has no common sense. Break it down to small enough pieces and write in very clear direct language, and they'll be able to follow it. I still left in. The goal is to entice readers to listen to this episode because we need the a So let's see what it did. What could it possibly do when you give it a template, an example, on the PDF of robust prompt and the episode transcript? Let's take a look. Here it is. You could see just by looking at it, and I'm gonna read it out to you, but it actually looks really close to the transcript.

Dan Sanchez [00:09:02]:
It has the 3 crucial parallels, and then it has 1, 2, 3, and then it wraps it up. And there's my call to action at the end. Oh, I have the quotes in there probably because I included them in the example. Oops. I could just take those out and run it again. It still probably is a little bit too wordy and long, so I could come back and try to recraft the prompt in order to deal with it better or make my example a little bit more concise execute closer to what you're hoping to achieve. The more clear you can make the end state that you want AI to accomplish, the more likely it is to actually get there or at least get close enough to where it's, like, 95% of the way there. So I hope you like looking at the step method.

Dan Sanchez [00:09:50]:
Practice it a few times. Try to make your own template. Go on a Google search and find your own examples and templates. Reverse engineer a famous creator's post and see if you can turn it into a template. It cannot work really well. There's a lot of different places and, places you can go in order to find these templates and examples where to go off of. Now stay tuned for the next part. If you think this is a deep prompt, wait until we go into the next one where I'm gonna be breaking down how to do multi step prompts in order to get AI to do even better.

Dan Sanchez [00:10:19]:
And we're gonna continue along the lines of building out this one LinkedIn post to be even better and better. So stay tuned for the next one in chain prompting.