Leveraging AI

74 | No code Automation made Izzy: The exact Blueprint that quadrupled Isabella Bedoya's business in 1 year

March 26, 2024 Isar Meitis, Isabella Bedoya Season 1 Episode 74
Leveraging AI
74 | No code Automation made Izzy: The exact Blueprint that quadrupled Isabella Bedoya's business in 1 year
Show Notes Transcript

In this session of Leveraging AI, Isar Meitis talks with Isabella "Izzy" Bedoya, Founder of MarketingPros.ai, as she makes her return since her appearance in our 17th episode. We'll talk about business automation with a blend of wisdom, and actionable insights that only Izzy can deliver.

Discover how to:
- pinpoint the perfect automation opportunities that will set your business apart,
- learn the art of prioritizing with a strategic mindset, and
- get hands-on with building your automation blueprint.

Izzy will share the tools she uses and reveals her step-by-step process to developing robust automation systems. This comeback is not just a webinar; it's an interactive experience where you'll gain invaluable skills to elevate your business operations through AI Automation.

About Leveraging AI

If you’ve enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!

Isar Meitis:

Hello everyone. And welcome to a live episode of the leveraging AI podcast, the podcast that shares practical ethical ways to leverage AI, to improve efficiency, grow your business and advance your career. This is Isar Meitis, your host, and this is a very, Interesting and fascinating episode. First of all, because I literally just made it to New York, 15 minutes ago and I had to get into an office, we work and find a space and set everything up, but I'm ready and we're good to go. So I'm happy about that. But I'm a lot more excited about the topic of this episode. So most of my episodes are very practical, tactical ways on how you should leverage AI in your business. And I host Amazing experts that share their skills and what they're doing with AI. Today's guest Izzy Bedoya is the only person that is coming on the show twice. So she's the first one ever to come in for the second time. That tells you how much I trust what she's doing, but also the other reason I'm very excited about today's show is a lot of you are asking me on LinkedIn and when you're chatting with me or people who meet me in conferences that I get to speak a lot on, like the one I'm doing in New York today and people ask me for actual use cases. yes, these things that you're sharing are amazing, but give me some hard numbers, like what results are people achieving with the processes that you're sharing? And so I lucked out because we scheduled this, I don't know, a month ago, but Izzy yesterday or this morning posted about her results with using this system. So here are some hardcore numbers for you on what using automation together with AI can do for your business. So in 2022, Izzy had a team of 12 plus people and plus, and when I assume plus, and I'll ask you in a minute, Izzy, I assume that's like subcontractors and people who are not actually core to the team. And she went down to a team of three. So that's a 75 percent or even more downgrade in the size of the team and her revenue tripled. So that tells you that this whole AI automation thing actually works. And if you see the trajectory and what she's doing in 2024, you will understand that it didn't stop in 2023, it just keeps on growing. And all of it, literally a hundred percent of it is built on AI. On building an automation machine that rhymes with a lot of AI capabilities in the background. If that's not exciting to you, I don't know what will be exciting to you. So hence why I'm really excited and humbled to have Izzy on the show again. Izzy, welcome to

Izzy Bedoya:

Leveraging AI. Thank you so much. Thank you for having me here. I'm excited for this conversation.

Isar Meitis:

So really, I, like I said, I lucked out. You literally just put out this post on 2022 versus 2023 and I think that would be a great starting point for people and to give people understanding what we're going to be talking about. We're going to be talking about exactly what Izzy did between 2022 and 2023 in detail. So you can do this in your business as well. So let's really start with how you got started, like what processes you had in your business and how are they working pre AI very quickly so then we can figure out what you actually automated and then we can dive into the how, which is going to be the majority of the episode.

Izzy Bedoya:

Yeah, absolutely. Absolutely. In a nutshell, excuse me, I'm just getting over a cold. but in a nutshell, basically, in 2022 and even since 2021, my main business has been like a Go HighLevel, agency, right? So I'm like reselling Go HighLevel, which is like a CRM, funnel builder, et cetera. And in the past, I had all these like little, like one off offers too, which created the need for more team and more. subcontractors and stuff like that. So one of the biggest things that I did is I just simplified everything and just focused on that one core offer. And then, we'll probably get into this, like along today. But, one of the biggest things that I did was I actually automated the entire client fulfillment process. So we could actually sell the SAS at a higher price point and still be able to deliver results for clients in a automated, process.

Isar Meitis:

Cool. Really, you had a SaaS slash service business around it that is now mostly automated if you have. So again, to put things in perspective, you're talking about having 50 plus clients a month and we are building like onboarded and served by a team of three. Yeah. So again, just putting things in perspective, that's not. Standard. Let's put it this way. I can use other adjectives, but I'll be very subtle with what I'm saying. It's incredible. So let's really go to what processes you started with. And you know what, I'll actually rephrase the question. What processes should a person start with? Like, how do you pick the best processes to start with AI automation? If you run a business and again, you're doing everything in your business, you're doing marketing, you're doing, product development. prospecting, onboarding, sales support. You have a lot of backend documentation that needs to be updated all the time because you're giving it to your clients as a way to use the systems and processes you're giving them and probably resell and upsell. So literally everything a business has, how do you pick? The first thing to start with, because if you look at the whole board, you're like, holy crap, I don't even know where to start. It's so scary. So I'm not going to do anything, which I'm sure a lot of people are in that situation right now.

Izzy Bedoya:

Yeah, that's a great question. And typically I think the go to that most people go to is, wanting to monetize or wanting to automate their, the sales process, which makes sense, right? You want to automate the sales process. You can generate more revenue. I think for this time around though, what I did is something different. I actually started with client fulfillment because I realized that, if I do want to grow a business, the biggest bottleneck is going to be client fulfillment. So if I start growing the sales and start bringing in a bunch of clients and we don't have the client fulfillment streamlined, it's going to be a nightmare. And that's pretty much what happened in the past year when I was actually having to manage a bunch of people and everything. we just didn't have a streamlined client fulfillment. That was one of the biggest headaches.

Isar Meitis:

Okay. So by the way, I agree with you a hundred percent. I think onboarding clients to disappointment. is a really good recipe for failure in the longterm because you're going to get really bad reviews and bad press and you won't be able to grow your business. So client fulfillment could mean a lot of things for a lot of people, but let's stay within the realm of what we're talking about right now, which is SaaS and service versus products, which I'm sure becomes a lot more complicated. So in this world of, okay, I'm delivering a SaaS product and I am providing a service around my SaaS product. What are the things you built as automations with AI in order to do this client fulfillment?

Izzy Bedoya:

So pretty much the entire, client fulfillment journey. So when a client comes in, we build out their funnels and we build out their, for example, we build out their bootcamp funnel or like a virtual event funnel. We build out all the automations for that funnel as well. Like the email marketing, the SMS reminders, like all of the automations that you could think of also tracking the lead from. opt into putting them in a pipeline and seeing the journey. So it was really like, that was essentially the offer. So then I had to figure out how to automate that whole thing. And I was able to do it inside of Go HighLevel and also using, custom GPTs. So now whenever someone comes in. We have a video that shows them how to, and it's all automated, right? So they get an email and it's Hey, welcome. You're in step one. Here's the link to, for the GPT. Here's the link for the video. And then here is, a link where you can actually submit your answers in this form. And when they submit their answers in the form, which the GPT gives them exactly what they need to put in. the reason why we didn't automate the GPT transferring to the form is so that the clients could have a little bit of freedom to update their, their copy before it gets on the form. When it gets on the form, then it automatically gets populated on the funnels. And then now I have Juan, who's like a member on our team, Juan goes in and just makes sure that the pictures are good and the branding colors are good and everything, so we're able to build funnels really like super fast and when they fill out the form it also auto populates their emails, their sms, it moves, like it just literally just does everything, so it allows us to, to help clients faster. Now one thing that I didn't do and I should have done it, But we just helped one of our clients do this, the actual videos, I actually filmed those videos, but we just had one of our clients do an entire course from end to end using, AI, like an AI avatar and AI voice and Canva for the, for

Isar Meitis:

the slides. Okay, so let's break this down for people, right? So the very first step that you must do is to map client journey, right? You need to actually know all the steps. It's not Oh, we know, like we talk to them on the phone, they do the thing, like you have to know the steps in a very detailed way and document it very clearly. Once you have that, you really can start breaking this down and how you automate each and every one of the processes. So those of you who don't know what Go HighLevel is, it's. It's like a one system does it all thing, right? It's a CRM, it's an email automation, it's a landing page platform. It's does everything you need in order to engage with prospects and customers and, but you can do this with other platforms, right? It doesn't really matter. It's the fact that you have these functionalities and really what you're saying is you built the entire onboarding process from including all the how to's. So for every step you put the client in addition to an email, there's a video, there's a process, so people cannot get lost. And like you're saying, you're feeding them with, with everything ready. Like you don't expect them to understand what to do. You're like, okay, copy this from here, put it in this thing there, and then you will get the results you need to get. So let's talk specifically about. the combination of GPTs and how this works with, because on one hand you have an email automation. I think everybody knows what that is and how it works. On the other hand, you have a GPT. I'm like, okay, those of you who don't know, GPTs are like mini customized versions of chat GPT that are. tailored to do something very specific. You give them instructions and they can do those instructions time and time again, very consistently, with or without specific inputs from the user. So you can tell it as part of the instructions, ask the user for X, and then it will give it to you and so on. So how do you connect the two? what are the touch points and how did you develop that whole system?

Izzy Bedoya:

So it was a lot of, It took me like four iterations to get to this stage. Every time it was just figuring out, one step further, one step further, how can I make this process smoother? And one of the things that I realized was that, this happens pretty much in like any coaching program, especially people join, they don't take action, they don't do anything. And then you want them to take action. so then I figured, okay, maybe it's just because they don't know. You know how to log in or whatever the case is. And, so I really had to like, think about how to simplify it and how to like really dummy it down to the point where it's like, they know, cause it's three components, right? It's a GPT, it's a form and it's a video. How do they know at any given moment what resources to use and all of that? and then I just thought about, how could I like super simplify it so there's no room for error. And then in addition, as soon as someone, like if someone doesn't fill out the form, it also has an automated reminder, like every three or four days to saying, Hey, don't forget to fill this form. So it pushes them to take the action step. but really it was just a lot of experimentation and seeing where people were getting like hung up and just also testing and trying different things, to see like, how could we simplify it just so that we can make the whole process super streamlined.

Isar Meitis:

So the form is in the GPTs. I'm just trying to make sure I understand.

Izzy Bedoya:

Yeah. so I programmed the, so essentially like the prompt that I use for the GPT, it's like, however many steps, like four, four steps usually, which is Target audience and then figuring out the actual let's say it's a boot camp So let's figure out the topic for the boot camp figure out the landing page copy figure out the emails Figure out the presentation slides like all of that And and specifically for the landing page copy I have it formatted so it gives us the landing page copy in a certain format So that we can just copy and paste that into, or they can copy and paste that into the form that they're supposed to submit. And then that form, it's called custom values. So that form has custom values that are programmed on the funnel already. So the form talks to the funnel and it just automatically populates.

Isar Meitis:

So I'm still confused and I apologize. It's probably me. Where does the GPT come into all of this? Like the form

Izzy Bedoya:

is still The GPT is the information

Isar Meitis:

provider. Oh, okay. Okay. Okay. So you basically using the GPT as a onboarding assistant to help the user figure

Izzy Bedoya:

out what to do. Yeah, it's like a coach hybrid with a copywriter. so instead of me having to sit with each person one by one, telling them how to create their offer, how to do everything, it's all programmed inside of the GPT, so it's basically like I have these GPTs that are acting as like mini coaches slash copywriters or whatever it is that they're programmed to do.

Isar Meitis:

And every step has its own GPT. So they're custom for that very specific task. Yep. Exactly. Now I understand. first of all, I think this is brilliant, right? So going back to what we said before, it's a multi step process. Like any onboarding of any client is a multi step process. And you have to support the users through that process, or they will fail, which means you will have less clients, or you will need to invest more resources in getting happy clients. And what easy did is basically create GPTs that help in the onboarding of people into the different steps and literally tell them what to do and what to copy and where to put it in and how to follow the different steps. Can you show us an example of one of those? So we can, instead of imagine, Yeah, you want me to share my screen? Yeah, sure, let's start with step one and then we'll see how much time we have, but they will be awesome to actually go through

Izzy Bedoya:

all the steps. All right, cool. let me go in here into the automation. I'll show you the, the whole workflow essentially, but it's essentially taking someone from start to finish. one of the, what we do basically, like I was saying is, as we set up like the funnels and the automations and stuff like that, so for my client, from, From an internal perspective, I should say when a client comes in, they get hit with an automation, right? And in this case, if we just skip over to like step three is let's build the lead magnet funnel. The first step is going to actually move them into the stage so we can have a visual in the pipeline to see the client, the lead magnet, then we'll get, an email that goes out and it's Hey, in this step, we're going to work on your lead magnet. This is the reason behind why we want to do the lead magnet. Here's the video. Here's the GPT to use. And then here is the, the actual form that you're going to fill out. So if I just put

Isar Meitis:

three, I want to pause it just for one second to, in order to explain to people, what is it that you're selling? Because I think that will make a lot more sense. So Izzy is basically selling the same system that she's using. Yeah. So everything that she is telling you, okay, create your own form, create your own onboarding, create your own landing page, create your own copy is because she's basically selling the same solution that she's using herself. And so when she's talking about this, what do you mean my funnel? That's what we mean. But if you generalize this could be explaining to a client, Any onboarding process, right? It doesn't have to be what is easy to do this for. This could have been anything. This is basically step one. What you need to do to get onboarded to the system. I'll give an example from my world. I just, I'm speaking at an event in next week and it's a very large corporation and I had to speak with five different people and go through three different systems to become a supplier to them. And so this is basically what this is trying to prevent, right? It's me self serving through an automation combined with a GPT that will help me through the process.

Izzy Bedoya:

Yeah, exactly. Exactly. and like you said, you can do, you can program this for anything. This is just like our client journey. for anyone that doesn't know what a lead magnet is, it's basically a funnel. It's like a landing page where someone can give you their contact information in exchange for a free download. So essentially that's what we're trying to build. So the email that the client's getting whenever they hit this stage and As you'll see, each stage essentially takes them somewhere else. in this particular stage, we're just telling them, we're just telling them, please fill out the lead magnet funnel. So it's telling them, this is a Loom video that's showing how to access the video in the portal. because believe it or not, that was one of the questions that we kept getting, like, How do I access the course or what video in the course, right? And it's like little things like that create hangups for people. So then when they go actually in the course, they go to build. And now here they have, we're talking about the lead magnet, right? So step three, and then in the lead magnet, they're going to see all of this information. They're going to see an example of how I use the custom GPT. They're going to see all this information, why they need the lead magnet. Now we named our, we named our GPTs to have names. So it has that Oh, you're working with people effect. So in this case, we gave this, Jimmy is a lead magnet one, And Jimmy is the one that's going to be responsible for creating your entire lead magnet from end to end. From the actual file that they download from the actual emails that got sent out the actual copy for the, for the thing. And as you can see here, I can just put, Or I should have, that should say Jimmy, but anyways, as you can see here, it's going to ask us a couple of questions. I'm just going to say create a mock scenario just for the, just for these purposes. So essentially the client has to answer these four things, and once they answer these four things, now they don't have to think from scratch. The GPT is going to actually create all of this information for them. And, And it's going to give you all of this, information. So as you can see here, the funnel landing page, this is now where it's formatted is specifically how we need it. So if you go back to step number four, that they're supposed to take here is fill out this actual form. And, Sorry, not that one. That's a revision. This form here. And as you can see here, it says, who is your target audience? And here it says target audience. So you just copy this part. And it says title, excuse me. So then it says title here. Then you just copy the title here. Then it says compelling headline and compelling headline. So you're just copy and pasting. Now, the reason why we didn't automate this part, That it's pretty much just filling this out. Reason why we didn't automate that is because, we want people to have that like flexibility of updating whatever was created by the GPT and then actually giving us something that is, what something that they would actually do. And now in terms of the actual, funnel, now the funnel in this case,

Isar Meitis:

Let's see here. So I want to pause you just for that, just one second for the people who are listening and not watching the screen. So what we're actually looking at is there are two components. One of it is a form that the users have to fill up in order to, Continue down the path of developing the solution that they want to develop, but it's hard. Like it's questions that they may or may not have answers to. And this is obviously a friction point that may cause them to either drop and not use it at all or ask for a refund or just bother your customer service because I need the help. I don't really know how to fill this thing out. And what the GPT does. Is it walks them through the process? It asks them simple questions that they know how to answer and in return gives them exactly what they need to put in the form. And if you think about any service provider for sure has this kind of an onboarding process, myself included. So I provide consulting to companies. I send them a form to fill out to understand stuff about their company. Sometimes. They know sometimes they don't know. So developing a GPT that can help your customers answer the question that are being asked in an easier, simpler way and generating the responses for them reduces a lot of friction, increases the chances of them doing it successfully without asking for help. And again, that's true for, Any client onboarding on the planet, probably, and it's just simplifying the process. And what Easy said about fully automation, in theory, Easy could have made the GPT write the data straight into the form without asking the user anything in between. But the idea here is to really give, like Easy said, instead of them staring at a blank sheet of paper, Here's a starting point. Is this good enough? Is this exactly what you need? Is this exactly what you want? Is this really your target audience or whatever else may be in your world? And then you can start with what the GPT gives you and continue from there to develop the idea or to add points. And it gives you a starting point, which dramatically simplifies the process.

Izzy Bedoya:

Absolutely. And, and that's actually one of the biggest things. And I think the stat is like 90 percent of online courses don't even get opened. and we were seeing that too, like when I didn't have this whole, automated onboarding process, I started, realizing a lot of people just don't know what to do, or they get too GPT really helps and it can, it creates the whole copy. And then here it says, Now that we've done this, the next step is actually, so it's, if I just open up this, this GPT, I can show you here that it actually has the steps. It actually has, step, step three, for example, after you generate the copy. Create the actual copy for the for the lead magnet and after that write a five part sequence So it's essentially everything that a course would have taught them how to do is just compressed into one gpt So it even saves their time our entire course Takes maybe about less than 90 minutes to go through all the videos and, but the work that we're doing, we're saving six months of work for any founder that's, setting up their funnels and we're able to condense it. And the fastest we've done a setup for is we had a client that he literally spent an entire Saturday. He said it took about eight hours. Spent that entire Saturday, filled out all the forms, did everything, and we were able to get him set up like super fast within a couple of days. So normally if you do this the manual route, you have to hire VAs and do all these different things and it's usually about a six month, Workload, right? So we're able to accelerate that. And then the funnel, this is like how the funnel looks, it looks like gibberish because these are all custom values that are, attached to this, the output of the form. So whenever they fill out the form, it's not, it's going to actually, fill out all of this information. So

Isar Meitis:

even this, so again, for those of you listening and not watching, what we're looking at is a framework of a landing page, right? It's just a wire frame. That's the word I was looking for, not framework. Yeah. A wire frame of a landing page that has. Multiple segments, like everything you need to have in the landing page. But instead of having the actual text and the actual content, it has these parameters in like squiggly brackets. But what happens behind the scenes is the automation, which I assume is built in GoHighLevel, right? So that's what actually fills up this form. Everything's in GoHighLevel. So the parameters that people, so again, I'll explain the process. So people understand there's an onboarding process. The onboarding process has a GPT that walks the user on how to fill up the form that will build everything they need in a faster way. Once they fill up the form, the automation behind the scenes. Actually builds landing pages, emails, and so on, on behalf of the user so they can start marketing to their clients with the minimal amount of work, which is obviously very attractive to anyone who is trying to create a marketing automation process.

Izzy Bedoya:

For sure. Yeah, that's the whole idea. and, and yeah, that's essentially what we were able to do. We just did this across. Thanks. three or four different funnel structures, so like a lead magnet, a low ticket offer, roundtables, and high ticket offers. And it's the same process. They just, watch a video, use the signed GPT, and then fill out whatever, form they need to fill out. And then also for revisions, just to make it easier for us as well, we streamline revisions so that they send us everything in one place. And it, it cuts out a lot of the back and forth. Um, so yeah, the whole thing is very streamlined.

Isar Meitis:

So I want to touch on two points and then I want to ask a follow up question. The two points are, again, if we generalize this to any universe in business, what a process like this enables you to do. Is to allow people who have zero technical skills to set up really sophisticated things because what's happening here is, and those of you don't come from the marketing world, take it to any other process you do know, whether it's a logistical process or an installation process or a setup process, whatever the case may be. But these processes require You to understand how the tools work. In this case, Go HighLevel. You need to know how to create landing pages. You need to know how to create email automations. You need to know how to define funnels. You need to know how to tie them all together. What Izzy and her team did circumvents all of that. You don't need to know anything. All you need is to answer questions in a form that you're being helped to do by a GPT and then everything else happens automatically behind the scenes, which explains what we said in the beginning, going from three clients a month with 12 people to go and do 50 clients a month with three people, because everything else just happens automatically. Now, so that's what I wanted to comment, which again, I find absolutely brilliant. I want to go back to the questions that I asked, because I think now people have a much better understanding of, okay, what are we talking about here? How do I actually do this? Like you sat down now, this, And you've been through iterations and so on. How do you start? what was the first thing, not that you did, because I'm sure you have a lot of bruises to say that you did the wrong things the first time around. If you recommend to somebody who doesn't want to use your system or want to use it for something completely different, what is the thing that you would recommend to me to do in my business tomorrow in order to build something like this for my business?

Izzy Bedoya:

Yeah, I think, that's a good question. I think the first thing is like really figure out where are your bottlenecks. for me, that was it, right? The biggest thing was the client fulfillment, but figure out where and actually sit down and audit your time. Like, where are you spending a lot of time? where are you spending a lot of time? What, Where are you putting a lot of resources, or people, or just investing a lot of money, figure out what that is, and then, just start there. it's the little wins that's gonna keep you motivated. I think for me, it was, like I said, I think it was first figuring out clan fulfillment, but, Depending on your offer and depending on what you're selling also, maybe it's not starting with client fulfillment. Maybe it actually truly is starting with, automation on like appointment setting or whatever it is. One of the other things that we did is, when we do virtual events, so because of the pandemic and everything, virtual events got really popular, but after the pandemic, there was a decline in virtual events. And one of the things that we did. As for us, it's always been really profitable and everything. So one of the things that we did in the last bootcamp that I did, I put an AI agent to call. So it was a three day bootcamp. On the first day, they just got email and SMS reminders. On the second day, they got email, SMS reminders, but they also had an AI agent call. Call their phones about 15 minutes prior to the actual event. So the whole reason was to like, get them to, see, it was more of a customer support call. Hey, just making sure that you got the link to the replay from day one. just curious, what are you here for? doing a little bit of qualification and then at the same time, also saying are you going to be in the call later today? I'm going to show you how to use. how to set up AI agents just like myself. So don't miss it. So just by doing that, it increased our show up rate from day one, which was like 40 percent of the opted in people, to 63 percent just because we had an AI agent call and get them a little bit hyped up for the actual call. and in the process, we got customer research out of it and stuff like that as well.

Isar Meitis:

Incredible. Two questions. One, which platform are you using for the call agent?

Izzy Bedoya:

So for the calls, I'm a licensed air partner. I can share a link probably a little bit later, but, air is really good. They're they've been the first to market in the last quarter, like last year, they had a little bit of a bumpy start, but because they're first to market, they're also like, The most advanced out there right now. There's other people like simple talk, bland, they're all great. but I'm seeing that a lot of them are having like the issues that air was having last quarter, they're having it now and it's, again, it's because they're just like newer technologies, right? So at the end of these, all these tools are, they're still, they're going to work great, you're going to encounter bugs here and there, and it's just because it's new technology.

Isar Meitis:

Yeah. So again, to explain to people, there are more and more tools out there that will call people and they sound like normal people. They're having an actual conversation. The delays are getting shorter and shorter to the point. It's almost indistinguishable between that and a human. And if the call is very short to the point and simple and very well defined, like in this particular case, Hey, we're having the call in 15 minutes. Don't miss it. By the way, what are you in for? And you even can feed that into some kind of a summary that will be available to the person running the meeting. Then he also knows like what people are coming in for and so on. So the ability to do that exists right now and whether it's perfect or not. And I agree with you, error are probably the best out there for simple things. There are better tool. If you really want to run like a full call center kind of thing, there are better tools out there. If anybody wants that. reach out to me on LinkedIn. I know a few people who actually run these kinds of companies, but, it's another tool in the toolbox that was not available to us before, unless we had people in a call center, whether offshore or here in the U S the ability to actually call people either a pre or post an event to remind or follow up is now available. For a ridiculous amount of money. Like it's ridiculous small. compared to now I need to hire six people to make a call. A hundred phone calls to make sure that people show up, to an event. So this is my comment. My next question after you did that is what was the, how long did it took you to figure out what this thing should say? Like the agent that, because you just script the beginning, right? And then it moves forward from there, but how did you figure out, or was it as simple as we're doing this event? Here's, what's going to be cool about it. Come join us.

Izzy Bedoya:

Yeah, like in terms of the AI agent,

Isar Meitis:

or which, yeah, in terms of the AI call agent.

Izzy Bedoya:

Yeah. Yeah. Yeah. So the, if you're using air, it does actually write a script for you, but I tweaked it. Of course. I wanted it to, you have to go through a lot of role playing with the agent as well. I wanted to keep it like short and sweet so it basically just says, if you'd like I can also show you but it pretty much just says, thanks for, joining the bootcamp just wanna make sure, I think it says something along the lines of like just wanna make sure, How you're liking it so far. Is there anything specifically that you wanted to learn? and then if they're an established business, it does a little bit of discovery. So that was just a lot of it was just, testing it just sitting there and having it call me and having these like conversations and, Me fighting back with it, saying I don't have time, don't call me and, seeing what it says and, I had to call my mom and my mom was like, she was like, Oh, do you speak Spanish? And then she starts talking to it in Spanish. It starts talking back in Spanish. So it's just seeing all these like different use cases happen, in real time. So I think that's been like the most important thing is just make sure that you spend time actually role playing with it. but yeah, if you don't

Isar Meitis:

know how to script, you're actually gonna be scripting. I'll extend what you said beyond that. Like all these things, when you build them in the beginning, they have issues. Like even the funnels themselves, like any automation thing you put together will break. And it will break for sure. The trick is to test 20, 30 different use cases yourself, fix the kinks you can find. So then it's going to break 1 percent of the time instead of 15 percent of the time. And it's going to allow you to onboard just more clients because you will address a lot of the issues in advance. And then when stuff happens, just find out what happened and fix that thing. And it happens to us still today, like in our automation, Oh, How did this user get this email? what the hell just happened? And then you go back and you find like a very specific scenario that you never thought of, on how things might happen. okay. so what you're saying, going back to my very first question is look for stuff. Takes you the most amount of resources to do right now, but that is repetitive slash tedious that you need to do all the time and start automating that. So I think it's a great point. answer. But then how do you figure out how to automate it? So let's try to make it generic and not your use case. I have a process that I'm trying to automate and it's the thing that's bugging me the most. And I have four people working on this every single week, half their time. How do I get started in figuring out How to automate it and what's going to be the best tools to do that and so on because you obviously figured it out. So what's the thinking process, not behind the specific automation, but on how to get started automating things.

Izzy Bedoya:

So you have to create like a roadmap, right? So when you audit your time, you audit what you're doing, you just have to actually document from point A to point B, what are all the little things you did in the process. And then once you have that clear picture, you first, you're going to see Oh, I could simplify this so much more, You're going to have that, that aha moment of like, how do I simplify it? And then once you're in that like setting, You can then figure out, Oh, I have to send an email. How can I automate that? And then that's where the curiosity kicks in because then you go on like Zapier, you go on make, or if you're using Go HighLevel, it has its own built in Zapier type of automation thing, right? So whatever tool you're using for automation. You just go in there and you just start seeing, can I automate this or can I automate, when I get an email response or when I, when, one of the things that, that, I was trying to figure out was, customer support questions, right? So it's like, what is the process? They send us an email, or they send me a LinkedIn or they send me a, They send me through all these different channels a question and then the questions sometimes fall through the cracks because I have to go through all the channels to find the client's question. So inside of the GoHighLevelAutomation section, I found that you could When you submit a form, you could send that information to a Google Sheets and then the Google Sheets can then send me a reminder that, hey, a question was submitted. So just by seeing that bird's eye view of like how your processes, then you train your clients on if you are a client and you have a question, this is where you ask it. Otherwise, I can't guarantee that you'll get a response. So for your own best benefit. go through this form. And, so you create your automation and then you train your clients and you train people how to use it so that it only makes you more efficient.

Isar Meitis:

That's a very good advice. And I'll generalize it again and make it very practical for people. So one of the things that I give to people who go through my course, and we have an AI, business transformation course where it's four sessions, and they're very detailed on how to take a business. On how to use AI across multiple things and how to build a strategy around using AI in your business. And very, it's a mix of strategy and tactics, right? it's both these things, but on the tactical side, one of the things that connects to what you do, we have a Google sheets. form that we've already created that allows you to fill in every task that you're doing as you're doing it. And then for each and every one, and it's, and it falls into categories. This could be finance. It could be marketing sales, HR, like whatever, department in the business has dropped down menus for all of these. And you pick the specific task you're working on. And really you can give this to every single employee in the company. And all they have to do is click the drop down menu and say, okay, I'm working on this. I'm working on this. I'm working on this. And within two weeks, you have a list of all the tasks that are done in your business, according to the different departments. And then for each and every one of these tasks, there are like. five or six or seven, I don't remember, drop down menus that basically ask you the questions that is related to, is it repetitive? Is it what's the risk of this process working not correctly? what is it going to do? How many hours a week or a month are invested in this process and so on and so forth. And by selecting the drop down menus, It gives this task a score of how effective it's going to be to automate it with AI or without AI. So there's actually two columns because people are now so now, Oh my God, I can do this. I'm like, there's a lot of stuff you can do with no AI, like just pure automation. Like a lot of stuff that you said, the form fills out the landing page information. That's. There's no AI in the process. It's just simple old school automation. So it gives you grades, basically numbers. And then you can see what's the things that you should probably automate first, basically, but to what you said, the things that takes the most amount of time in your process. The other thing that you can add to that is really what's on your critical path of stuff you're doing in your business, right? There could be stuff that's taking a lot of time, but it's not going to. Generate a lot of revenue. It's going to be not going to save you a lot of expenses, just stuff you have to do, then maybe it's less urgent than the other things. So there's ways to really map the processes. The other thing that we recommend to people to do. And again, I'll recommend it here. And I apologize. I'm taking time from you, but I, but it's, I think it's a great brainstorming session for both of us, to drill down to the process itself, like to the level of, okay, what are you doing when you're doing this task? You can ask people. Okay. To record themselves on the screen, if it's a, in front of a computer thing, or to record themselves on their phone, if it's a field kind of task, it's not, I'm not sitting in front of a computer and then you can transcribe what they're doing and if it's a screen recording, you can grab the screen recording as well, and then you can load that to any, your favorite large language model, Claude, ChatGPT, Gemini, like whatever you'd like to use and ask it to build an SOP out of it based on what people are saying that they're doing. And then it knows the time it's taking because it knows the length of the recording. It knows exactly what the person did, which screen they're going to. And all they're doing is doing what they're doing anyway, sharing it in words. I'm clicking on this, I'm opening this menu, I'm going to this thing, I'm typing that, I'm copying it from here to here. And it can build an SOP from you. And you can use this as a baseline to the automation. So there's multiple ways, even if you have nothing, to build the initial map. So now you can start figuring out what to automate and how to automate. Yeah,

Izzy Bedoya:

I think that's the best way. It's having that like full roadmap.

Isar Meitis:

So I got one more question on the practical side of the GPTs because I find it really fascinating the way you did it. And again, those of you who didn't build a GPT, don't be afraid of it. It's literally like using chat GPT, only telling it what will happen if whatever different scenarios. One thing is again, how did you get started? Did you get started like on a document on the side with the flow chart? Like, how did you get started programming it? And second, how did you test it out? did you actually play the client again and again and again?

Izzy Bedoya:

Yeah, the mock, yeah, the mock up scenario. it was a lot of testing. probably like this has been, and I know earlier you said, I'd probably learned a lot and I have a lot of bruises through it all. So I've had to recreate this program maybe like four times in the last, I don't know, six months or so. and every time has just been, Automating and making it more efficient every step of the way. So I think by doing it so many times, it just was natural. what are the steps that the client has to take? so now I thought about that and I put it in a GPT. if you want, I can share it with you. My, my little template thing. Yeah. That'll awesome. So a lead magnet. Lemme

Isar Meitis:

float up because I was just trying to create a very detailed long GPT. And first of all, I found out it's limited to 8, 000 characters or words. I don't remember. I think it's 8, 000 words, which I did not know, but I hit the limit. I'm like, Oh, I did not know these things have a limit. So now you have to go back and simplify your instructions because you literally run out of room. And I'm like, I was able to do 6, 000 words instead of 8, 000. okay, I should have started with being it more, accurate in what I'm asking for. So I'm sure again, I, it will be awesome to look at your process.

Izzy Bedoya:

Yeah, and that happened to me too. I also found when I was putting stuff in like the knowledge base It wasn't like performing as well. alright, so this is like my custom GPT instructions, basically This is like this is the backbone of it all and then I just changed the instructions here So just to go through it I start by saying what is your goal? what is the role? So you're going to be like a lead magnet specialist or whatever. And then I just remind them of the actual outcome that we're looking for. And then, and then this is just like all best practices in prompt engineering. You want to give it like a raise. You're going to tell it to breathe. You're going to tell it to, you're going to put if you don't do this, you're going to get punished kind of thing. And then all the, like sticks

Isar Meitis:

and carrots.

Izzy Bedoya:

This is all you don't touch any of this. You just go and you just change this and this. now the, anything in the exact instructions. So pretty much here, this means that this is where you actually share your input, your, steps. Now, the reason why it's like in these like carrot things is because it's pretty much saying like this prompt here is saying don't share anything inside the exact instructions.

Isar Meitis:

which by the way, I assume, is not really bulletproof, like people can still get to it, but yes, I do the same thing, even

Izzy Bedoya:

though I know, yeah, there's some people that they do a lot of throw a lot of gibberish at it and it'll eventually crack, but, but for the most part, I have tried cracking these GPTs and I still haven't been able to do it. So anyways, the, what you want to do is you want to put your step by step instructions. So for example, in the, In the GPT for, here, let's just use this one for example. In this one, I'm saying like, okay, there's four types of posts that you're gonna create. You're gonna create videos, carousels, freebie, and relatable posts. And then I give it like exact instructions for the video post. You're gonna do this for the, and then here it's creating the script, right? And then you have what's the next one after that? After that is the carousel. So it's the actual format for the carousel. So it's just very strategic, step by step instructions. I don't, even though it's AI and it's intelligent and all of this, I don't give it a lot of room for it to go off on its own tangent because I want to be very method, like very controlling of the output. because I'm very controlling of the output, then I'm giving it like very. Controlled instructions, so it can only generate information in the way that I want it so that I can copy and paste it really easily into whatever, To whatever I'm doing and same here freebie post it's gonna auto generate so I'm gonna say create a mock scenario Using the freebie post template. I'm just gonna skip all the instructions. Yeah, and and it should just generate Here, just using all the stuff in squiggly brackets is essentially, it's essentially what is generating. So as you can see, now you have something that looks a little bit more like a LinkedIn style post, and then you just have to do minor tweaks. so yeah, that's how I do it. I have my base prompt here, and then I just put my exact step by step instructions that I wanted to do. And as detailed as possible. Now, if you don't really care for the output, if you're not trying to be controlling of the output that's generated. You can keep the instructions a lot more, you can keep it a lot more generalized, but for my purposes, I have to be very controlling of the output because it's going to make everything else so much easier. In our stuff.

Isar Meitis:

Yeah, absolutely. I listen, I want to say two things and then I can, I think we can sum it up because you and I can probably geek about this for another week and I'm sure you have other stuff to do and the listeners as well, but the one thing that's very clear and that relates to like the tips I gave before, you have to be highly organized to do this. Like before you get started, you need to know exactly what the outcome needs to be, what the format of the outcome needs to be, what are the steps that will allow you and, or the bot to get to that outcome in a very way, what formats are required, what data as an input you want to get either from a third party or from the user, et cetera, et cetera, et cetera. So the first thing you need to do is to be. Very good at mapping out the outcome, the inputs and the outputs and the process in between. And then you can go and start building these things using either the format that easy shared or another format. But again, also having a structured format that you know that works is going to save you time. So it's like a prompt library for creating GPTs, that works very well. Every time you need to create another one, I'll generalize something else. We have been creating a lot of GPTs for ourselves. And literally in the process of just you said, so we're not using it with clients, but we're using it. Like every time we, there's something we do regularly, we're like, Oh, let's just build the GPT for it. And then we do. And then that's it. You, and then yes, we fix it in the next three to four weeks. And then six months later, you fix it again. It's a very helpful way to slowly, but surely remove tedious tasks from different teams in your company. Easy. This was nothing short of incredible. I think what you're doing is amazing. If people want to follow you, find you, learn from you, take your bootcamp, work with you, what are the best ways to do that?

Izzy Bedoya:

First of all, yeah, thank you for having me. I'm really honored to be here, for a second time. best place to follow me is just on LinkedIn. I, I also started a YouTube channel. I'm posting a lot of tutorials. and if you do want to try out the AI agent, if you go to my feature, there's a link where you can download it. I included a tutorial, so it's really easy for you to set up and get started with your own AI agent. and it's a free download. so definitely check that out.

Isar Meitis:

Awesome, easy. Thank you so much. This was absolutely fantastic. I'm sure there's going to be a third time and you're probably going to be the first one to do the third time as well. So I appreciate you. I appreciate what you're doing. If you guys want anything like this, go check it out. There's really nobody else out there that does it better than Izzy. Thank you. Thank you. so much.