Atiba De Souza / Ai Expert. CEO Strategist. Team Productivity Expert

Atiba De Souza is your secret weapon when your team is underperforming. 

As a CEO strategist and team productivity expert, Atiba transforms struggling organizations into high-performing machines using AI-powered systems and proven leadership frameworks. 

In this electrifying episode, Atiba breaks down the difference between AI tools, agents, and employees—and why every business leader needs to rethink their org chart NOW. 

From his 1996 basement-built search engine to today’s AI employee revolution, Atiba’s journey is packed with eye-opening insights, practical steps, and a few headsmack moments that just might change the way you lead.

GUEST BIO: 

Atiba De Souza is a CEO strategist, team productivity expert, and AI innovation pioneer. With a background in software engineering dating back to the 1990s—when he built and sold a search engine from his mom’s basement—Atiba has spent decades optimizing performance in both government and private sectors. Today, he helps businesses build AI-powered systems that enhance productivity, improve decision-making, and bridge the gap between human leadership and digital execution. He’s also the author of The Delegation Trap, and a dynamic voice on the intersection of leadership and technology.

Link: clientattractionpros.com

Paul Povolni, the founder of Voppa Creative, has been a creative leader for over 30 years, with clients around the world. He’s led teams in creating award-winning branding and design as well as equipping his clients to lead with Clarity, Creativity and Culture.

Headsmack Website

Transcript

Paul Povolni (01:52.962)
Hey, welcome to the Headsmack podcast. name is Paul Povolni and I am excited to have another misfit with me. I have Atiba D'Souza and he is your secret weapon when your team is underperforming. As a CEO, strategist and team productivity expert, he transforms struggling organizations into productivity powerhouses through proven leadership frameworks that deliver immediate results. Atiba, how are you doing,

Atiba De Souza (02:17.671)
I am Greatpa, how are you my friend?

Paul Povolni (02:19.944)
I am doing awesome, man. I am liking the Superman swag that you have on your head. I want to hear the story of how that became your thing. I am very much into Superman representing here with my t-shirt. Noah got it in the background. Yeah, yeah, yeah. That's my thing. So are you excited about the Superman movie coming up?

Atiba De Souza (02:33.243)
I know it and I'm loving that black and go the guy back in gray. Yeah, I'm loving that black and yeah, that's beautiful.

Atiba De Souza (02:45.041)
You know, first and foremost, I'm excited about all superhero movies. And so, yes.

Paul Povolni (02:48.686)
You're my type of person. I am a superhero fan. I'm not a superhero nerd or a geek, but I love superhero stuff. I'll watch anything superhero. so looking forward to it. So what do you think? Do you think it's going to be a good one, bad one?

Atiba De Souza (02:57.241)
Me neither.

Atiba De Souza (03:06.811)
It's going to be interesting, isn't it? So this is this is the thing for me with DC. DC has an opportunity right now because quite honestly, Marvel after Avengers has been OK. Right, like they haven't put out anything that really made people go, yes, and you're back in and you're super excited. You know, Black Widow was good and yeah, it's OK. Right. And so I think they have a real opportunity here.

Paul Povolni (03:23.148)
Yeah. Yeah.

Atiba De Souza (03:37.105)
to recapture so many of our attention for an appetite for really, really great superhero story. And I mean, who better than Superman for that? Yeah.

Paul Povolni (03:48.418)
Yeah, yeah, the original, right? Yeah, I'm pretty excited about it as well. I'm actually gonna be seeing it on the eighth on the IMAX. And so that is pretty exciting. I think it's a great cast. I've seen the cast in different things, just bits and pieces of them. I think the cast is absolutely perfect for it. I think that the Superman is a combination of both, you know, Smallville Superman and Henry Cavill. I think it's a great combo and

Atiba De Souza (04:07.516)
Yeah.

Paul Povolni (04:17.654)
Yeah, I'm excited about it, man. I'm pretty pumped, but we are not going to be talking superheroes today. I do want to hear a little bit about your origin story, You know, tell me a little bit about where Atiba started. You can go as far back as possible, but I just want to hear a little bit about your origin story and learn more about you.

Atiba De Souza (04:36.093)
Yeah, absolutely, Paul. So I've been well, let me just start here. 1996, let's go all the way back there. OK, pre 2000. So 1996, myself and two friends who had been working together already for a couple of years doing some projects and stuff. We sat in my mother's basement at the time, because that's when all good things, even your mom's basement or your garage, like all good things happen. And yeah, in the 90s. And we.

Paul Povolni (04:43.735)
Yeah, let's do it.

Paul Povolni (04:58.958)
Yeah, in the 90s,

Atiba De Souza (05:05.5)
We built a search engine in 1996. And we had a search engine online in 1996. Now, we did not know what we really had. So much so, Paul, that the next year in 1997, we sold it for a whopping $2,500.

Paul Povolni (05:08.205)
Wow.

Paul Povolni (05:22.604)
Wow. Wow. Wow.

Atiba De Souza (05:24.997)
Yeah, so I missed the boat by a few zeros. Yeah, like seven.

Paul Povolni (05:30.518)
Yeah, yeah, yeah, my goodness.

Atiba De Souza (05:33.756)
Right? And it's so funny because shortly thereafter, after we sold ours, and we were excited, I remember my mother coming downstairs and saying, hey, have y'all heard of, and I think it was, it either Excite or Alta Fista, one of those really, really early ones. And we're like, no, we never heard of it. We said, well, look it up. And we looked it up, and they had just sold for something like $5 million. And so we were sitting with our 2,500, and they had $5 million. And we were like, hmm.

Paul Povolni (05:57.326)
No.

Paul Povolni (06:02.558)
Wow. Wow. Now where did the skill come from for building a search engine? Like that's not, that's not an easy thing. Where did that come from?

Atiba De Souza (06:04.273)
Yeah, so we learned really quick.

Atiba De Souza (06:10.225)
Yeah, so me and these two friends were true nerds and geeks back in the day. So today, I have to wear the Superman hat so I can look cool. But I'm really just a nerd. And so we had been building, our first software application was in 1992, where the three of us got together and built an online casino.

Paul Povolni (06:19.534)
But underneath your Clark Kent.

Paul Povolni (06:32.174)
Well.

Atiba De Souza (06:38.375)
Well, wasn't online. It was a computer-based casino. I say online because of today, but it was a computer-based casino, right? And it was just a thing that we did and we had fun. We didn't make any money off or anything, but it was just something that we did. And we kept playing around and we started building websites and doing other things. And so we were at a place where we had all of this information and all of these bookmarks. And if you remember back in 96, mean, bookmarks suck today. They were awful back then, right? In Netscape.

Paul Povolni (07:01.838)
Yeah, yeah. Right.

Yeah. Some people just don't know the pain of the nineties internet.

Atiba De Souza (07:07.901)
Right? they don't. And so that's how we ended up building a search engine because we realized we had all of these thousands of bookmarks and we needed a way to find the information that we wanted quicker.

Paul Povolni (07:27.426)
Wow. Wow.

Atiba De Souza (07:30.205)
OK? And that's what we built. And so that's how we ended up in that. From there, that started me in search. From there, we ended up doing and building search systems for the government, big systems for the US government. Then came back private sector. And that was about 2005 when we came back private sector. And at that time, Google was in a quandary. And the quandary Google was in was

Paul Povolni (07:32.312)
Yeah.

Atiba De Souza (08:00.827)
There were so many really smart marketers who were figuring out how to game the system and put junk at the top of Google search. Right?

Paul Povolni (08:08.652)
Right, White lettering and all of that stuff.

Atiba De Souza (08:12.605)
all the black hat tricks that they were using. And so Google had a problem. Now, because of my background in search, I understood how Google was going to solve the problem. And I said, well, if I understand how Google is going to solve the problem, and now we're going to have all these business owners who need the problem solved, that's what really kicked off the SEO agency that we've had for many, many years. Did that for a very, very long time.

through that process, mean, I've built, bought, sold all kinds of businesses and had all kinds of experiences, still own an agency today. And today, so I own an agency and a lot of my focus now is on teaching a lot of the stuff that I've learned over the years, lot of the hit smack stuff.

Paul Povolni (08:49.443)
Yeah.

Paul Povolni (09:02.22)
Yeah. Yeah. Now, did you ever find out what happened to the people that bought your search engine for 2,500 bucks? Did they ever do anything with it or did they miss an opportunity as well?

Atiba De Souza (09:15.591)
They missed it too because and part of that was the way we sold it to them. And us not even understanding that the bigger opportunity that it was at plan. I mean, it wasn't their business. You know, they weren't a software company. We ended up selling it to them so that because they had a back then a really big website with a hundred and something pages and people could never find anything. And so we sold them the software as site search.

Paul Povolni (09:40.983)
Yeah.

Paul Povolni (09:45.422)
wow, yeah, yeah.

Atiba De Souza (09:46.981)
OK, I'm still friends with the owner of that company. And it's still in the way back machine. If you go back and look at it, it's still there. It doesn't work anymore, but it's still there what we built all those years ago. Yeah.

Paul Povolni (09:51.171)
Yeah.

Hahaha

Paul Povolni (10:00.3)
Yeah. So, so that led you to, you know, building and selling and growing your company into what you do now. So, so what did, when, when did the transition go from, you know, the, building of software to what you're doing now? Are you still building software?

Atiba De Souza (10:18.065)
So interestingly enough, my agency is back in the software building arena. But now we're building AI employees that support your marketing efforts.

Paul Povolni (10:30.942)
wow. So, so that's an interesting thing. I think people that might be listening to this has had probably heard the term, heard the phrase, but don't really fully understand what that actually means. Like, is it really an employee that can it really replace, you know, certain types of people? So let's, let's dive deep into that. because I know that's, that's where your, your genius is. And, you know, I got to speak at an event that you also spoke at and heard some of your genius there and

Atiba De Souza (10:47.697)
Yeah.

Yeah.

Paul Povolni (10:59.404)
So the Superman had and saw your vibe and I'm like, okay, we got to connect. so I do want to hear about, you know, what does that mean? What does it mean to have an AI employee and what does that look like? for people and then, you know, transitioning to who should be worried and who shouldn't be worried about it.

Atiba De Souza (11:04.109)
Absolutely.

Atiba De Souza (11:12.027)
Yeah.

Atiba De Souza (11:18.525)
Well, and that second part is the fun part of the conversation So here's the deal. The reality is the term AI employee is an evolutionary term In other words, it's changing daily in terms of what its definition is Right and they're different levels to this. So, you know Was it a year and a half or so ago a year ago? Open AI came out with custom GPTs where you can write little bits of code and it did something

Paul Povolni (11:22.092)
Yeah.

Atiba De Souza (11:47.737)
And some people call that employees. Those aren't really employees. Those are specific tool-based tasks that do a thing. They do a thing. Now, I don't know about you, but I don't know anyone who hires an employee to do one single thing. Okay? Even if it's one process, there are multiple things in that process. And GPTs only typically do one sliver piece of that. Okay? All right, so let's hold that thought for a moment.

Paul Povolni (12:04.376)
Yeah.

Atiba De Souza (12:17.485)
Then we started coming up with this idea of the AI agent. Now the agent says, OK, I can do a thing, but I can also think about the thing that I'm doing. In other words, I can make some reasoning choices, and I can also have a bit of a workflow and a process, wrapping all of those together into one.

Paul Povolni (12:32.236)
wow.

Atiba De Souza (12:46.043)
That becomes an AI agent. Now, AI agents are cool. So for example, an AI agent can go out and say, manage your Google Calendar.

Paul Povolni (13:00.3)
Wow. Wow.

Atiba De Souza (13:00.647)
Right? And that's cool. It can be smart enough to realize that you just requested to book a time that someone else has already booked for and then say, hey, you can't book this time. Do you want to move this appointment or what do you want to do? Great. That's an agent. An employee now is multiple agents working together.

Paul Povolni (13:22.957)
Yeah.

Atiba De Souza (13:30.215)
So now an employee can say, hey, I can manage your calendar. I can also do emails for you. can also, you've got a meeting coming up with ATIBA. I can go out and research ATIBA for you and create a document for you about ATIBA so you can be prepared for your meeting. you already met with ATIBA once before and you have notes in say your CRM.

Paul Povolni (13:30.424)
Wow.

Atiba De Souza (13:56.657)
can go to your CRM, pull those notes, synthesize it all together.

Paul Povolni (14:01.43)
Wow. So, so let's, before we can, I mean, this is, this is mind blowing for me, but also I think for those listening. So when we talk about even building an agent, before we talk about building an employee is an agent, is it a special software? Is it special using of existing software? Like, like where does an agent come from? Like what, what, you know, you, you, you use that phrase, you know, and, I think by now people are familiar with.

Atiba De Souza (14:02.46)
Okay.

Atiba De Souza (14:13.969)
Yeah.

Atiba De Souza (14:22.129)
Yeah, how? Yes.

Paul Povolni (14:30.048)
GPTs and custom GPTs because they've been around for a little longer. So it's kind of become a part of our language. but agents is, is still very new for some people. So where does an agent come from? Where does, where does that manifest? Yeah, exactly. Right. That's, that's exactly what I was doing. We are too, too much alike. My mind was totally when I kept saying the word agent, I'm totally thinking of a black suit, white, you know, white shirt, black tie.

Atiba De Souza (14:32.945)
Yes.

Yes.

Atiba De Souza (14:44.175)
Welcome to the Matrix.

Atiba De Souza (14:51.303)
You

Atiba De Souza (14:56.678)
Yes.

Atiba De Souza (15:00.327)
But that's kind of what it is. And quite honestly, when the term was coined, that's what they were thinking about. They were thinking about the matrix. So you're spot on. So the answer to your question is multi-fold. You can pull up a notepad in your computer and write code that will create an agent.

Paul Povolni (15:01.794)
Okay.

Paul Povolni (15:09.548)
Wow. Wow.

Atiba De Souza (15:30.421)
That's the old school software way. There are programming languages that you can use, Python and so on and so forth, to write agents and create agents. Most of us can't do that. Hell, I've got a background in software engineering and I don't even want to do

Paul Povolni (15:48.526)
Yeah.

Atiba De Souza (15:49.637)
Okay, so that's one way. Now that is the purest way. Now what we have today are no code tools. So what no code tools are is saying, okay, the code still has to be written, but instead we're gonna put a pretty front on it that the average user can use and say, okay, do this, do this, do this. Kinda like when WissyWig came out.

When we were doing websites, what you see is what you get and you can drag and drop. It's the same thing for code now. Okay? And so you don't have to know the code. Now you just need to know how to use these different blocks and put the different blocks together. Okay? Still takes logic. Right? And there are tons of no code tools that exist. Everything from cursor to DeFi to N8N.

Paul Povolni (16:19.842)
Yeah. Yeah.

Right, right.

Atiba De Souza (16:47.791)
Etc etc etc and I mean there are diamond dozen and they keep coming out All the time and it really depends on your level of proficiency Okay, cuz some of them are truly made for front-end users who have never done anything in software before But they're also very limited

Paul Povolni (16:52.684)
Right, right.

Paul Povolni (17:07.459)
Yeah.

Right.

Atiba De Souza (17:12.227)
Others are still no code, but they're pretty close to the line of code.

Paul Povolni (17:16.91)
So let's talk about the one that is limited, even though it is limited, but for somebody that's like, man, some of the stuff that a T-Bus shared already about the calendar stuff and all of that, that is pretty, pretty snazzy. want to, I want to do some of that stuff. So what are some of those tools? Because I understand what you're talking about that. It's basically when you do this, then you do this, then you do this, right? That's what you're talking about. These little nodes of.

Atiba De Souza (17:41.233)
Yeah.

Paul Povolni (17:43.242)
of behaviors is, know, when this happens, do this behavior. When this happens, do this behavior. So, what are some of those tools if somebody was to want to say, okay, I want to build an agent and I don't want to get too complicated. I'm not a code person. What are some of those tools like and what does that look like for somebody to even build something super basic?

Atiba De Souza (17:47.249)
Yeah.

Atiba De Souza (17:57.788)
Yeah.

Atiba De Souza (18:04.545)
So both that new is one that a lot of people use. There's another one that's slipping my mind right now, and hopefully it will come back before we're done. We don't use any of those, which is why I don't know them off the top of my head, to be completely honest with y'all, right? Because they're too, for us at least, they're too basic, right? Now, here's the other side. You have things like Zapier and Make.com.

Paul Povolni (18:16.974)
Yeah, yeah, yeah.

Paul Povolni (18:25.569)
Yeah, okay.

Paul Povolni (18:32.481)
Right.

Atiba De Souza (18:32.495)
which are automation tools. So remember we talked about custom GPTs and all that type of stuff. Automation tools are even before that. An automation tool is you tell it exactly the process to go through.

Okay, and it's step to step to step to step and you define the entire thing.

Paul Povolni (18:49.069)
Right.

Paul Povolni (18:52.834)
Right, right.

Atiba De Souza (18:54.307)
What Macon Zapier have done is that, okay, let's take our automations and allow you to insert intelligence in the middle of an automation to create an agent.

Paul Povolni (19:06.614)
Wow, okay.

Atiba De Souza (19:08.249)
Okay. Versus like there's a there's an agent that that we released just yesterday actually. And maybe as business owners this might be helpful. You know how situations come up in business with staff with employees and sometimes you just don't know how to respond. But you know you have to respond and how you respond is going to be very very important. And

Paul Povolni (19:37.405)
Right, right.

Atiba De Souza (19:38.223)
And sometimes the response you want is the one that you know you probably shouldn't give.

Paul Povolni (19:44.238)
Just like posting on Facebook, right? Your first thought shouldn't be the final thought.

Atiba De Souza (19:51.779)
Exactly. Right. And so we built a tool that actually allows us as business owners to come and say, Hey, here's my situation here to stakeholders, the different players in the situation. Here's how I want to respond. And here's the outcome that I need to have. And what it does is it actually walks you through from your current thinking into the thinking and the questions that you need to ask in order to get to the outcome that you want.

Paul Povolni (20:08.206)
Wow.

Paul Povolni (20:19.266)
Wow. Wow. So that's, that's almost like having a, once again, a personal employee. That's like a, I guess a little, a little coach or a little devil and a little angel on your shoulder. And when one says, don't say this and the other one says, say this instead. Yeah. Yeah.

Atiba De Souza (20:27.206)
A coach.

Yeah. Yeah. Say that. Yes. And so that's an agent where we built that and there is no logical process. All right. We just built an agent inside of DeFi to be exact. And it's a chat agent because you actually talk to it back and forth with the voice agent. Right. So you tell it your problem and it responds to you in a conversational tone.

Okay. And so with that, that's just an agent that we said, Hey, here's what we need you to do. Here's what we need you to think about. Go figure out how to do it. And it does it.

Paul Povolni (21:12.834)
Wow. Wow. So we've, we've talked about a couple of different, different ways to use and, I guess they're different forms of agents. know, we talked about the make or this APR where you, you do a behavior when I do this behavior, you do this behavior and then inserted into that is some intelligence, right? And so that's kind of a, a very, different agent to what you're talking about where it's a conversational agent, which is different to the one that you talked about where

Atiba De Souza (21:15.11)
Right.

Paul Povolni (21:42.835)
It's a manager's your calendar and your CRM and some of that stuff. Am I getting that right? Is that okay?

Atiba De Souza (21:49.979)
Yes, those are all different ways of thinking about it. so Paul, one of the things that we teach people and we share is when you're looking at your org chart, because that's what it all comes back to when we're talking about AI employees, whatever definition that we're using. When you look at your org chart and you're planning it all out and you're saying, you know, we have a deficiency here and you're thinking, well, how can AI help us here?

right, wherever that may be on the org chart. The key is to look at it from one out of three aspects and break that position down into one of these three aspects. What is the strategic work that needs to be done for this position? What is the tactical work that needs to be done for this position? And what is the analytical work that needs to be done for this position? When you look at it,

like that, it's a lot easier to see where AI can help.

Paul Povolni (22:55.575)
Wow.

Atiba De Souza (22:57.947)
And so let's say AI could do all three. I, in that case, would build three, at least three different agents to make up that one employee.

Paul Povolni (23:08.704)
Yeah. Now I noticed you didn't have knowledge as one of those three things because knowledge is not a thing when it comes to AI, right?

Atiba De Souza (23:16.283)
Well, yeah, knowledge is kind of ubiquitous. It's there, and you know everything needs knowledge. And it's in the background. And so.

Paul Povolni (23:19.02)
Yeah. Yeah.

Atiba De Souza (23:28.443)
Here's the beauty of the beauty of knowledge. When you build your agents correctly, they learn and they grow and they get better. They teach themselves, which I know that sounds a little scary.

Paul Povolni (23:41.582)
That sounds terrifying, especially when you talk on agents, because then we're moving from the Matrix to the Terminator, right? And so that's where it gets scary.

Atiba De Souza (23:52.581)
It does a little bit. It does a little bit. And this is why, for me, there always still has to be human oversight at this juncture. And for the sheer fact, so this is, here's the uncomfortable truth that nobody's willing to tell you about AI and AI employees, but we've all experienced it, but somehow we ignore it. Have you ever run the exact same query?

Paul Povolni (24:02.626)
Yeah. Yeah.

Atiba De Souza (24:22.639)
inside of chat GPT and gotten two different answers.

Paul Povolni (24:26.722)
Yeah. Yeah. Yeah.

Atiba De Souza (24:27.663)
It happens. happens to all of us. Guess what? It happens to agents too.

Paul Povolni (24:33.236)
Hmm... Well...

Atiba De Souza (24:35.047)
So I'll give you an example. So I have an executive assistant, AI agent, and it sends me my calendar every day, right? And today it sent me my calendar and I knew that I had this meeting with you today, but it didn't send that to me. And I'm like, why? Why did it miss that? And it always tells, here's your calendar for, and it gives me the date, right? And so this message comes back. Here's your calendar for undefined date.

And then it was missing stuff and I'm like, what the hell happened?

Paul Povolni (25:08.558)
Hmm.

Atiba De Souza (25:10.801)
Why did you do it differently this time? And so that is also part of the challenge here with agents, which is very different than software. So when we were doing software development, we programmed it so that it always did the same thing every single time. And it wasn't variance. Variance was bad.

Paul Povolni (25:33.336)
Right, right. It was zeros and ones.

Atiba De Souza (25:36.625)
whereas hair variance is every day.

Paul Povolni (25:40.162)
Wow. So how do you fix something like that?

Atiba De Souza (25:44.67)
That's why human has to be in the loop. In this case, happened. We haven't figured out why it's the undefined yet. I didn't even look into that yet. But with your calendar, was because of the way the meeting was scheduled. It's on your calendar, but it's not actually on my calendar as owned by my calendar. And that's where the problem came in. So something in the way we scheduled caused it to trip up.

Paul Povolni (26:08.593)
okay.

Atiba De Souza (26:15.335)
But these are the things that you don't realize, think about why it did what it did.

Paul Povolni (26:21.282)
Right.

Atiba De Souza (26:23.173)
Right? So yeah, these things happen. But that's why human has to stay in the loop. And so as much as people are terrified of it learning and it becoming sentient and it going off, we build human breakpoints in the loop of things to make sure, yep, this is right.

Paul Povolni (26:29.614)
Yeah.

Paul Povolni (26:44.45)
Yeah. Yeah. I check some balances, making sure things are right. Now I loved what you shared about looking at the org chart and looking at those three things that you need to consider when it comes to an AI employee. So let's talk a little bit deeper about that. As far as, you what, and you could even like dissect those three things even deeper if you want to.

Atiba De Souza (27:00.145)
Yeah.

Paul Povolni (27:07.278)
But talk a little bit more about those three things when it comes to a business. Cause you know, you're very much about productivity and making sure things are as efficient as they need to be. know, finding the glitches, the things that aren't as streamlined as they need to be and fixing that. So talk a little bit about those and dig a little deeper and unwrap those a little bit more.

Atiba De Souza (27:24.892)
Yeah.

Absolutely. let's not make this hypothetical. Let's think about your org chart for a moment, What's a position, what's a role that's missing on your org chart that you would like to fill? Or one that you have filled, but the person who's there hasn't done everything. There's still pieces that are missing.

Paul Povolni (27:50.828)
Well, I'm a solopreneur, so I might not be the best thing. So we might need to find another example,

Atiba De Souza (27:56.359)
But even still, have you ever had a contractor or thought about maybe bringing in anyone else?

Paul Povolni (28:02.478)
yeah, definitely.

Atiba De Souza (28:03.793)
So who would you bring in?

Paul Povolni (28:05.478)
well, let's talk about the podcast. I'd bring somebody in to, to, to create this podcast, to edit it and then to distribute it on social media.

Atiba De Souza (28:14.811)
Okay, fantastic. Okay, so now with that position, the first thing is, is there strategy that's involved in doing this job?

Thinking about it, it's pretty much a tactical job.

Right. And so maybe not so much strategy. OK. So we might skip strategy for a because it's not very strategic. OK. So it's a tactical job. All right. Now it's a tactical job. so tactically, when we started and I hope I'm not giving away too many secrets. When we when we started, you started a recording and then you asked me to say my name and the podcast name three times. OK.

Paul Povolni (28:54.382)
Hahaha.

Atiba De Souza (29:07.291)
Then there was a little bit of a pause and then you started the recording or you started the actual podcast and we've been talking and at some point we will end, right? And it will be an ending. Okay, that's the recording. Now, do you do an intro?

Paul Povolni (29:18.285)
Yeah.

Paul Povolni (29:26.848)
I have one pre-created, yeah. The intro is usually you introducing yourself, and then I go into the music, and then I go into the recording.

Atiba De Souza (29:29.053)
Okay, yeah.

Atiba De Souza (29:34.973)
perfect. And so you have me introducing myself, the music, go into the recording, then you do an outro.

Atiba De Souza (29:49.149)
What other alucia?

Paul Povolni (29:50.23)
You froze up. Yep. You're back.

Atiba De Souza (30:01.191)
Huh?

Atiba De Souza (30:07.717)
recording.

Atiba De Souza (30:16.551)
I don't know if you can hear me. I'm typing in the chat.

Paul Povolni (30:17.294)
Hello, hello.

Paul Povolni (30:22.766)
Hello, hello.

Paul Povolni (30:30.723)
Are you there?

Atiba De Souza (30:37.013)
I I lost a spot.

Paul Povolni (31:12.398)
you

Atiba De Souza (31:40.413)
Okay, well, pause since we're still recording, I'm gonna kind of finish where we were. Okay, so you're probably gonna do an outro maybe, and maybe you have some commercials that you want spliced into the recording for this podcast. So if you were gonna create an agent, a tactical agent now, that was going to work with your podcast, then what would need to happen is anytime, and I'm just kind of making this up here.

Okay, so let's say you are in Riverside and once you take the files, hey, welcome back.

Paul Povolni (32:19.097)
Hey, I'm back, man. I'm sorry about that.

Atiba De Souza (32:21.285)
I just this, no, no, it was fine. And I figured I just didn't know when you were coming back. I just started and kept going, going, but we can, we can do it together. And I guess you'll have to cut it, but actually we'll make a really good point for the agent. Okay. So the agent, right? Maybe you put an outro on the end. Maybe you put commercials in your podcast recording. Okay. Let's say you do any of that type of stuff. Fantastic. it's an audio and it's a video.

Paul Povolni (32:27.481)
All right, thank you.

Atiba De Souza (32:50.225)
Podcast and so eventually it needs to get to whatever podcasting service you use and onto YouTube Now in order to get there it needs a podcast description. It needs a title. It needs a thumbnail created It needs a transcript. It needs all this stuff. Okay fantastic So if we were looking at this we say, okay, we can create an agent that anytime

a recording is dropped into a Google Drive folder. So it leaves Riverside and you drop it in Google Drive, right? It was monitoring that Google Drive folder. It goes and it picks up the audio. It picks up the video from that folder. It transcribes them. So now I've got audio, video, and I got a transcript. Fantastic. Okay. Then it knows that you're going to do the first

section of the recording where it's kind of banded between you and the guest and then you have the guest say their name three times along with the podcast name three times. And so it knows, if you tell it you're gonna do that, it knows to listen for that and then there's gonna be a pause. And then it can cut that piece off and know the rest is the recording. Okay? Now it has your recording.

Paul Povolni (34:04.325)
Wow.

Atiba De Souza (34:15.835)
It has the transcript of your recording. can now go write your show notes. It can now go write your YouTube description. It can figure out a great title. It can use image generation prompts to then create a thumbnail for you. Then let's say you're connecting to Lisbon as your service. It can connect to Lisbon or whatever podcast service and post your podcast, connect to YouTube and post it to YouTube.

Paul Povolni (34:43.545)
Wow. My goodness. Wow.

Atiba De Souza (34:44.669)
There you go.

Okay, now human in the loop. Where I would insert human in the loop in there is clearly on this podcast, we had a little bit of a technical issue. And that technical issue probably needs to be cut out.

And so, I mean, so after it does all the editing, we may have it drop what it created into Airtable for review by a human. Now, human may watch it and say, hey, from minute X to minute Y, please cut.

Paul Povolni (35:17.115)
Hmm.

Atiba De Souza (35:25.713)
and then it will go back in, cut that piece out, put the video back together, drop it back into Airtable, human comes back in the loop and says approved, and then it goes and does the posting.

Paul Povolni (35:38.555)
Wow. Now, something that you mentioned there, and you've said it a couple of times, is human in the loop. I think that's a phrase that maybe people hadn't thought about when it comes to creating AI, AI agents, automations, and things like that, is that sometimes there is the need for the human in the loop. Now, the acronym, I guess, is HIDL, but that's not become a thing yet.

Atiba De Souza (35:44.21)
Yes.

Atiba De Souza (35:57.959)
Yes.

Atiba De Souza (36:01.789)
It has not.

Paul Povolni (36:02.491)
But that hasn't become a thing yet, but that's a necessary part in in creating These is figuring out when there needs to be a human in the loop

Atiba De Souza (36:11.643)
Yes, and think about it, if you hired a video editor and a podcast editor after they did the editing, you might review it. And so in normal employee interactions, we have human in the loop.

Paul Povolni (36:30.959)
Yeah, yeah, right.

Atiba De Souza (36:32.741)
Right. And so we can replace those and still use them in the AI examples. Right. So, yeah, and you can build as many in as you like. Obviously, the point here is to build as few as that are only the ones that are necessary. Right. Otherwise, you defeat the purpose. But we have. And human loop, depending on.

Paul Povolni (36:52.367)
Right, right, right.

Atiba De Souza (37:01.105)
the sensitivity of what you're changes. We have faceless YouTube channels that we manage and the human in the loop happens after the video was created and posted.

And so after it's posted, the human comes in and says, no, that wasn't a really good video that you created and then delete it.

Atiba De Souza (37:26.501)
Right? Because those channels are producing content every single day and I'm not going to have a human reviewing every single day.

Paul Povolni (37:35.033)
Ryan, Ryan.

Atiba De Souza (37:36.509)
So it just depends on your use case and the sensitivity of what you're creating.

Right?

Paul Povolni (37:44.923)
So going back to something that I had, a question that I had asked earlier, who should be nervous about this and who shouldn't be nervous about agents and especially AI employees? Who should be nervous and who shouldn't?

Atiba De Souza (37:58.418)
Yeah.

Atiba De Souza (38:02.437)
So number one, SaaS companies should be very nervous.

Paul Povolni (38:07.13)
Okay.

Atiba De Souza (38:07.631)
And SaaS companies should be very nervous because SaaS was great. It was a tool that allowed you to do other things. But you still had to go in and do the thing in the SaaS. And now as we have employees or agents that can do the thing that the SaaS was doing, it becomes, do I need SaaS? But for example,

Paul Povolni (38:18.862)
Yeah.

Paul Povolni (38:31.163)
Hmm.

Atiba De Souza (38:35.579)
I love CRMs, not gonna call any particular names, but I love CRMs. They're great. I've always loved CRMs. And now my agents are going out to the CRM to get data, to come back and do something with it. I'm starting and the agent has an entire data store of knowledge base of me so that it can act on my behalf with contacts.

So I'm starting to ask myself the question, so why not have all the data from the CRM with the data that's me? Why does have to be separate? Why do need the CRM anymore?

Paul Povolni (39:13.433)
Hmm. Wow. Yeah.

Atiba De Souza (39:16.957)
Where my agent can do all the things that the CRM does. Why do I need it anymore? Now, we're still a little bit off from completely just killing all of SAS. But I do see that as a really, really big thing that's coming. So that's one area, one sector that should be really nervous. Another one is anyone who's not willing to think.

Paul Povolni (39:22.394)
Yeah.

Paul Povolni (39:41.755)
Wow, yeah, yeah.

Atiba De Souza (39:44.215)
And so I had a conversation last weekend with a group of people in the education field. we're talking elementary and high school education. And they were flabbergasted by how many of their peers are just using AI and turning in lesson plans from AI that even say obvious things that you can tell it came from chat GPT.

Okay? Yeah, you're not thinking.

You're going to go the way of the dinosaurs. This life that we're going into is for the creators. Now, I don't mean the painters and the artists. I mean those who can think and create, those who can dream and want to bring their dreams to reality.

Paul Povolni (40:21.817)
Mmm, wow.

Atiba De Souza (40:40.721)
That's who this life is for. That's who this life is going to favor. And if that's not you, yes, you should be worried. You should. You're going to get left behind.

Paul Povolni (40:47.887)
Hmm.

Paul Povolni (40:53.973)
And that comes from being able to use AI, being willing to use AI, being highly skilled at using AI. When you talk about creators, who are you talking about?

Atiba De Souza (41:06.749)
I'm talking about every single human on the face of this earth. Okay? All right? I'll give you the story of how I first recognized this. It was December after ChatGPT came out, and it had just come out the month and a half before in October, and someone asked me to do a presentation to a group of kids in Africa. I forget exactly which country it was, but they had about 25 kids.

Paul Povolni (41:11.419)
Yeah, OK.

Atiba De Souza (41:37.025)
in this room in Africa and we're doing a Christmas presentation for them. And I'm talking about the power of artificial intelligence. These kids, and I'm talking, they were six years old to 18. These kids had never heard of chat GPT before that day.

Atiba De Souza (42:00.911)
In that presentation, in that hour, I was able to walk with those kids over Zoom a month and a half after ChatGPT came out and from the six-year-old up, help all of them define business ideas that they can bring to market.

Paul Povolni (42:23.001)
Wow.

Atiba De Souza (42:23.141)
just by exploring the things that they're curious about.

Paul Povolni (42:29.198)
Wow.

Atiba De Souza (42:29.425)
they can create. And so when I say every human on the planet, I literally mean it. And so those who have sat and always just waited, you're going to get left behind. Those who will say, I don't know if I want to adopt, you're going to get left behind. Because this train is moving way too fast.

Paul Povolni (42:44.793)
Yeah.

Atiba De Souza (42:54.747)
and it's not about to slow down.

Paul Povolni (42:58.125)
Right. And it's evolving so quickly. They can get very quickly left behind.

Atiba De Souza (43:01.191)
Yeah.

very quickly, because I teach a web class. we got it down now to, I'm only doing two topics. And we rotate once a month these two topics that I teach in the web class. And we got it that way because one of the web classes is how to perfect the content on your website, your landing page content that attracts your ideal customer. And every time we teach a class, month to two months later,

It's a brand new class because the technology has changed so much. gotta update it.

Paul Povolni (43:34.971)
Yeah, yeah, wow.

Atiba De Souza (43:40.045)
There's better ways to do it in just a two month span on that one little thing.

Paul Povolni (43:48.751)
And that's with using AI. Like what, what is, what has changed recently about those things that, that you're teaching.

Atiba De Souza (43:50.588)
Yeah, yeah.

Atiba De Souza (43:56.571)
So for example, the first time we did it, it was on, OK, how to take your content and use JAT GPT to create your ideal customer profile, use your ideal customer profile to then figure out your best landing page content based on your offer that you're offering, blah, blah, Great. Then notebook.lm came out. And then it became, so.

Here's now how you can first evaluate what your current website is saying to see the delta that you need to change as you then go build your ICP, have your offer, build the content, and then reevaluate it to see what is actually saying and how you're showing up. Then it became, forget all of that because now between chat to Gemini, back to Claude,

and then into lovable, you can actually build the entire landing page and be done in an hour.

Paul Povolni (45:00.601)
Wow.

That's insane.

Atiba De Souza (45:10.203)
And so in that first iteration, we were barely just getting you content on a page that says, here's the content you should use. Go find a developer who can hopefully change it for you. In the second one, we were able to show, here's the content, and here's why it works. And we gave you reasons why it would work. And so you felt really confident. And then in the third one, you walked away in an hour with an actual brand new website.

Paul Povolni (45:36.559)
Yeah, that's insane. And it's constantly evolving. And that's the thing that makes it such a fascinating thing to look into and to keep up with and to understand and to learn and try and experiment with and all of that stuff is because it keeps getting better and better. Now there are, as you mentioned, there are things where you do need a human in the loop, where you do need to check stuff and you can't just outright take it as

Atiba De Souza (46:01.991)
Yes.

Paul Povolni (46:06.683)
as always true because, know, there are sometimes glitches, there are sometimes things that, that I'm not quite correct. Right. And so, so for somebody that, that is wanting to dive into this stuff, and I do want to, you know, learn more about what you're doing and how you're helping businesses as well. for somebody that's wanting to dive into this, what do you recommend they start with? Do they start with GPTs and

Atiba De Souza (46:16.487)
Yes.

Paul Povolni (46:31.355)
Uh, you know, do they start with manners? Do they start like, where do they, should somebody start with if they wanted to kind of get their feet wet? What's the gateway drug to this?

Atiba De Souza (46:40.613)
Yeah, that's a really great question. And the truth is a lot of people start incorrectly because it's so cool and I just want to go run and do so much stuff. And you end up actually doing a lot of nothing. And the reason that that happens, and so this is a big part of what we do and what we teach, is everything that we do has to be principle based. It has to be.

Paul Povolni (46:52.217)
Ha ha ha.

Paul Povolni (46:55.962)
Yeah.

Atiba De Souza (47:08.273)
based in principles of business, has to be based in principles of communication, it has to be based in principles of human psychology. Hold that thought. So that's a really, really important thought. The next piece, and I love, I love all the people who are saying prompt engineering is dead and you don't need to know prompts anymore, you're idiots.

Okay, I said it and you can tell him I said it because here's the deal. Those of us who can ask better questions are going to get better results. Yes, are the models getting smarter? They are, but that doesn't mean we get dumber. Just think about it. If you have a better tool, why not get better at using it? Why are we regressing? They're not making it better so we can regress. That's not the point.

Paul Povolni (47:48.484)
Hmm.

Atiba De Souza (48:00.637)
That is not creation. OK? And so, yes, the place to start is learning prompt engineering and learning prompt engineering across the different models, whether that's OpenAI, whether that's Claude, Perplexity, Gemini, learning the differences between them, when you use what, and when you use them in combination with each other and how.

Paul Povolni (48:29.295)
Wow. Okay.

Atiba De Souza (48:29.841)
Because when you get to agents, agents need to understand that. you need, in agents, what we're doing is writing prompts. When we talked earlier about that question coach agent, it's a prompt. It's a three page prompt.

Paul Povolni (48:39.044)
Yeah.

Atiba De Souza (48:48.667)
And if you don't know how to write the prompt, the agent's not going to produce for you. So the basic fundamental, I know there are all the no code tools and you want to go build something, don't.

build the fundamentals first, the principles. And the principle is understanding how to build prompts and do prompt engineering. To that point, one of the other things that we're releasing this week is a prompt engineer coach, where you can actually go in and tell it, hey, this is what I'm trying to do. I'm using OpenAI. This is the outcome I want. And it will actually walk you through and then help you write the prompt.

Paul Povolni (49:30.746)
Wow.

Atiba De Souza (49:31.133)
It will ask you the questions like if you did, if you left stuff out, it will ask you the questions it needs in order for it to write a great prompt for you.

Paul Povolni (49:41.263)
That's amazing. So when it comes, you'd mentioned there's a difference between open AI Claude perplexity, grok, Gemini, you know, there seems to be so many of them. do you have like a brief, like a super brief rundown of what the differences is right now? Because I know early on they were kind of neck and neck. They were very similar, at least in people's minds. And then they've kind of diverged into specialties, right? Or am I getting that totally wrong?

Atiba De Souza (49:56.027)
Yeah.

Atiba De Souza (50:07.473)
Yeah. So you're not totally wrong. And it's worse than that. And so I will be brief. in this brevity, let me say the reality is you have the different companies. And then they have models underneath each of the companies. And each model underneath each of the companies does something different. So yeah.

Exactly. So the order of operations in the exponential, it gets more complex when you start looking at all of that. But let me give you the brief like this. OpenAI is great for ideation. You want to come up with ideas, go to OpenAI. OK? Or chat GPT, in this case. Right? Go to chat. Chat is great for that. Inside of chat, you have the

Paul Povolni (50:36.493)
Yeah.

Paul Povolni (50:43.405)
Yeah. Yeah.

Atiba De Souza (51:02.957)
Models that start with and in the models that start with numbers all the old Models are the reasoning models. Those are actually agents That are gonna think about what you ask it versus just going and getting you an answer and dumping it back Okay, that's open AI Google and Gemini and specifically Gemini 2.5 Pro Which I love that they call it pro and they give it to us for free. I love it. Thank you Google

Paul Povolni (51:31.867)
Nice.

Atiba De Souza (51:32.986)
Makes no sense, but that's what they do fine This is the strategy agent

When you want strategy, nothing does strategy better right now than Gemini 2.5 Pro, in my opinion. Claude is by far the best writer.

Atiba De Souza (51:57.521)
So often we may go to chat GPT to get ideas, to even get a skeleton to say, hey, I've written this part of it as well, incorporate all of this, and then we'll take it into Claude and say, Claude, now, hey, help me clean this up. And Claude is a great, great writer. Perplexity on the other hand is designed for research.

Paul Povolni (52:14.137)
Yeah.

Paul Povolni (52:19.557)
Hmm.

Atiba De Souza (52:28.803)
It wants to be the greatest question answer bot ever.

Paul Povolni (52:32.847)
Yeah. He wants to replace Google, right?

Atiba De Souza (52:34.32)
And so, yeah, so that's where you go. You go to Plexity for those types of questions, okay? We internally don't use Grok very much. Personally, I just think Grok is a little too cranky.

Paul Povolni (52:50.491)
Maybe it's in the name, it just sounds cranky.

Atiba De Souza (52:55.225)
And we haven't found, there are a lot of people who love it, and no issue whatsoever with that, right? Wherever you love it or don't. We just haven't found the use for it. In those four, we can do just about everything that we need. So we haven't had the need for it.

Paul Povolni (53:04.869)
Yeah.

Paul Povolni (53:11.481)
Yeah. Well, and in the way you described them too, it totally aligns or syncs up with what you had shared earlier about, you know, sometimes when you're creating those agents or those employees is you would go to different, specialties or skill sets, and then kind of bring it all together. And so you go to one first, they do their thing. And then, like you mentioned, you know, you might start with the idea generation in, in open AI, and then you take that into, now

Atiba De Souza (53:35.559)
Yep.

Paul Povolni (53:39.789)
articulate this in a better way, you know, through Claude or, you know, maybe there's a step in there is research this a lot better. You go into perplexity, right? And then funnel that out and make this into an understandable document or blog post or paper or whatever. And that would be through Claude, right?

Atiba De Souza (53:44.85)
Yep.

Atiba De Souza (53:50.055)
Yep.

Atiba De Souza (53:58.779)
Yes. And the beauty for us, and I'll show you how much for geek we are, geeky, we built an agent that can take an input and tell you what's the best tool to run this in.

Paul Povolni (54:07.569)
Hahaha

Paul Povolni (54:16.415)
there you go. Solve that too,

Atiba De Souza (54:17.733)
Right? Because sometimes you don't know. Yeah.

Paul Povolni (54:23.375)
That's amazing. So, so I want to hear a little bit about what you do. I mean, you've, you've kind of shared it already and it's just, it's just mind blowing in what you do. So, so how do you help companies with building these agents, building these employees? What kind of tools do you use? What kind do you offer and how do people get ahold of you?

Atiba De Souza (54:23.484)
Yeah

Atiba De Souza (54:42.845)
Yeah, it's a great question. Thanks for that. And thank you for listening to us all the way to the end, too, for this. So here's the deal. We're actually launching a brand new community. It's so brand new, I haven't even named it yet. But it's going to be launched by August 1, 2025. And in this community, what we're going to be doing is bringing together three things.

Paul Povolni (54:44.92)
No.

Atiba De Souza (55:09.563)
that I have not seen anyone else bring together before. You heard me say I'm very principled and everything has to be based in principles. So we're going to be bringing together marketing principles along with leadership and team building principles, along with AI as we learn to build AI employees all in one community. Now, as part of that community, you're going to get access to a lot of the AI employees that we've already built.

that you can just take and use on your own. And we're going to be building more. As you heard me say, we're releasing stuff every single week. Now, if you want to know about the community, because like I said, I have not even named it yet. OK, go to meetatiba.com. That's going to take you to my LinkedIn. Send me a message on LinkedIn and we'll get you an invite to the community. here's the other thing.

Paul Povolni (55:54.692)
you

Atiba De Souza (56:07.429)
I'm making this community extremely affordable and you have to apply to join. Because if you're just someone who wants to come in and take, if you're just someone who wants to say, AI, cool, gimme, gimme, gimme, gimme, and you're not aligned with us on principles and you're not aligned with us on the fact that this is collaboration, we're all learning together, this isn't the right place for you.

Paul Povolni (56:32.549)
Yeah.

And so what was that? What was that link again?

Atiba De Souza (56:38.203)
meet atiba.com.

Paul Povolni (56:41.913)
And so, so how else are you helping businesses, you know, with, with some of the, the, the services that you provide? Because for some people they're like, man, this is so awesome, but there's no way I've, I've got the brain for this stuff. I need somebody to help me out. Do you help people and how do you help them?

Atiba De Souza (57:00.635)
Yeah, absolutely. So two sides of that. One is we do have a marketing agency. And so if you're in a place where you're having a hard time attracting your ideal customer, we are an organic video marketing agency where we use video to attract your ideal customer to your door. And that's a lot of the agents that we're building here of doing all the parts of doing video marketing. And so that's one side. More of a done for you type of service. Then we also.

outside of the community have it done with you, where being part of the community, you have access that we can help you do implementation as

Paul Povolni (57:41.665)
Nice. Nice. Well, man, this has been an amazing conversation. I've absolutely loved this. Mind blown. I mean, I've heard a lot of these terms. haven't dug deep into it myself. I know if I do, will geek out and just like waste days playing around with this stuff. I'm just wired that way. My kids are wired that way as well. And they absolutely love AI and they've just dug deep. My son just built an app because he needed to solve some problem that he needed to solve.

Atiba De Souza (57:47.505)
Thank you, Paul.

Atiba De Souza (58:02.077)
It's awesome.

Paul Povolni (58:10.587)
And he literally built an app that solves a problem for him and automate some stuff for him. And so it is, it is the future. You can't ignore it. It is how things are going to happen for your business, for every business. And whether you're, you survive it depends on how you engage with it, adapt it, learn it, get familiar with it and all of that stuff. And so I do have a, a two questions before, before we wrap up.

Atiba De Souza (58:38.94)
Mm-hmm.

Paul Povolni (58:40.2)
The first question is why the Superman brand?

Atiba De Souza (58:44.069)
Great question. So as we talked about a little bit earlier, I love all superhero. And so I actually didn't choose Superman. What ended up happening was I coached football for a lot of years. And almost 13 years ago now, I took over a high school team. And I'm on my way to the first day of practice in August. And you know, it's super hot. got two a day practices and I forgot my hat at home. Well, you can't survive outside.

that long in the heat without a hat. And so I pulled into the local Walmart, like I just need to get a hat. I need a bucket hat. And they had the classic fisherman bucket hat with the mesh on top. And like that helps no one because it's still baking. Right. Or they had a Superman bucket hat. And so, yeah, I like well, Superman it is. And I end up wearing that hat the entire season. By the end of the season, some of the other coaches had started

Paul Povolni (59:27.545)
Ha ha ha!

Atiba De Souza (59:42.545)
buying me other Superman stuff. And that's where the Superman hat collection started. And so the brand chose me. I didn't actually choose it.

Paul Povolni (59:50.085)
Nice.

That's awesome, man. So the final question is, you know, I want to make sure that, you know, we've we've addressed or covered or talked about anything that is relevant for the discussion. So what's a question that you wish I'd asked you?

Atiba De Souza (01:00:10.547)
that's a great question.

So the one thing that we didn't cover much, which is perfectly fine too, is the interplay between human and AI employees and how that shows up, especially as a business owner and a leader in your organization.

Paul Povolni (01:00:33.477)
So talk about that for a second.

Atiba De Souza (01:00:35.323)
Sure, thanks for asking.

Atiba De Souza (01:00:43.229)
Again, as I said a little bit earlier, everything is about principles and it's about human psychology and the way people work and how we communicate. Right. And one of the things that we've learned in the last 18 months of building AI employees from GPTs all the way up to multifunction agents is the way we think about how we delegate to them is exactly the way we need to think about

how we delegate to actual humans.

Paul Povolni (01:01:15.547)
Hmm.

Atiba De Souza (01:01:17.111)
Nothing changes. actually wrote a book about this. Nothing changes. And that is crucial to understand because so many people are running to AI employees to get away from humans and thinking that if I go to AI employees, my life is going to be easier until that AI employee forgets or misses to put Paul on your schedule for today and you miss an employee.

a podcast interview, and then what do you do? You're going to get upset with the AI employee like you would the human? How do you react then?

Okay, now you're starting to see why I wrote that question, coach. Okay, and so we have the same situations. They're going to get things right. They're going to get things wrong. Sure, they're not going to take sick days. Sure, they're going to be more efficient, but they're also going to make bigger screw up mistakes.

Paul Povolni (01:01:56.293)
Yeah.

Atiba De Souza (01:02:15.709)
And when they do, there'll be no one to blame but you. And right now, when your staff makes a mistake, you blame them. But the truth is, it probably 99 % of the time was your fault that they made a mistake. Just like 100 % of time when the AI makes a mistake, it's your fault.

Paul Povolni (01:02:27.023)
Wow, yeah.

Atiba De Souza (01:02:41.789)
And until we learn the true principles of delegation, none of us should be bringing AI employees in. Just my opinion.

Paul Povolni (01:02:49.435)
Wow. Wow. Wow. Now, can we have AI employees check other AI employees? Like a quality control, like for that calendar thing, you know, would there, would there be a sequence where it says, you know, separate AI or separate agent or separate employee check this employee's work? Like that, that happens in business, right? You have people that proof things, you know, a copywriter proof this website copy that the marketing team came up with or whatever.

Atiba De Souza (01:02:55.931)
We do.

Yes, we do.

Atiba De Souza (01:03:07.973)
Yes. Yes.

Atiba De Souza (01:03:16.082)
Yes.

Paul Povolni (01:03:18.093)
So is that possible? Is that happening now?

Atiba De Souza (01:03:21.157)
Yes, so we have several AI employee systems that have feedback loops from other AI employee systems that know what you're supposed to be producing, get your output, checks it over, says, nope, that's not good enough. One of the ones, for example, is we have a headline generator. OK, now you know, writing copy, writing headlines is an art form. Right? And so, yes, there is an AI agent that

Paul Povolni (01:03:45.924)
Right?

Atiba De Souza (01:03:50.895)
understands the customer, understands the content, produces a headline, and then there is the feedback loop that checks what you produced and said, no, that's not good enough based on, and get this, this is where it gets fun, based on what we know has worked in the past, that's not good enough because of X, Y, and Z. Modify.

Paul Povolni (01:04:11.771)
Wow. Wow.

Atiba De Souza (01:04:15.099)
And then it modifies.

Paul Povolni (01:04:17.615)
That's amazing. That's amazing.

Atiba De Souza (01:04:18.757)
Yeah, absolutely there is supervision.

Paul Povolni (01:04:23.831)
Right, right. By other AIs, even if it's not human in the loop, it's just another AI that is coming from a different realm or a different box that's checking on this one because within that box, something might've gone amiss. so, cause you got to do that. I don't think we can a hundred percent trust AI for several reasons. There's also bias when it comes to AI.

But there's also sometimes you can just get off track. so, man, Atiba, this has been fantastic. You it's you've been mind blowing. I knew this would be a great interview when I saw your presentation and have absolutely loved it. I know you have you mentioned the book. You have a book coming out. And so if people reach out to you on LinkedIn to get a hold of that, I'm sure you're going to market it on there if they want to connect with you. Is that the best place to connect with you on LinkedIn?

Atiba De Souza (01:04:53.682)
Yeah.

Thank you.

Atiba De Souza (01:05:08.071)
Yes.

Atiba De Souza (01:05:18.831)
Absolutely. Especially if you want to have an actual conversation with me. That's the best place to go. Just be, and I want to have a conversation with you. so, you know, send me a DM on LinkedIn. Let's talk. That's what it's there for. Let's actually connect human to human and chat. Right. And yes, introduce you to the community. The book is TheDelegationTrap.com. And it's all about how we learn to delegate better.

Paul Povolni (01:05:23.225)
Hahaha

Paul Povolni (01:05:46.767)
Nice, nice, love that. And that's a whole nother discussion as well. So maybe we need to do a follow-up to this conversation. That's another head smack is the delegation. Yes, absolutely. So Atiba, this has been fantastic and appreciate you coming on and I've enjoyed this and I'm sure you've provided a lot of head smacks for people that have maybe never considered AI, have heard about the different things it does, but it never really understood as

Atiba De Souza (01:05:51.516)
Yeah.

Atiba De Souza (01:05:54.791)
That's another head smack.

Paul Povolni (01:06:17.349)
broad or as much as it can do. And so this has been very enlightening, man. Thank you so much.

Atiba De Souza (01:06:22.941)
Paul, thank you. Appreciate you, You too.