AI Agents Are Your New Employees (Here's How to Hire Them)

and potentially that task that took me 5 hours, but now it took me 30 minutes.

Welcome, we are here with Tim Kakir from AI Operator. He specializes in helping teams become AI confident. He's trained thousands of professionals and helped many teams just change the way that they work. So we're really going to start diving into some of the use cases and how you can apply AI immediately to solve those pain points in the business. So first things first, thanks for making some time, Tim.

Thank you very much for having me, David, or Dave. It's an absolute pleasure to be here. Thank you.

Yeah, perfect. Well, yeah, when you start working with someone, you were telling me just before we hit record about a little exercise that you do, and I thought that would be the perfect place to start. So which is, where do you get started with AI?

Yeah, I think that when we work with clients, the first thing they're like, oh, we want AI to do this incredible thing. And I'm like, have you done that before yourself? And they're like, no. And I said, why do you want AI to do this? Because AI can do it potentially. I'm like, okay, how many hours do you spend in reporting? You know, they're like, maybe 5 hours. And I'm like, how does your salespeople do follow-ups? And how long in, in like, what's the time to, time to answer, time to follow up? And they're like, well, it's 48 hours. And I'm like, oh, okay, well, there's so many other things that you should get AI to help you with first. And then we can look at all that sexy stuff that AI can do. So we have an exercise called It Sucks That. It's a human prompt. It's not a, don't put it out to AI, but you take a pen and paper. I think a pen and paper is one of the best things to do because we, you know, like we don't take pen and papers often. Like I did the other day. I was like, oh wow, this is awesome. It sucks that. It sucks that I need to report on Fridays. It sucks that I need to prepare pitches to investors. It sucks that I need to report to our investors. It sucks that my salespeople takes 48 hours to answer a lead. It sucks that, you know, you make a list of it sucks that, and then you'll see as you've written them down, you'll be like, oh, okay, well, why are we doing all these though? And as you put that in that paper, you'll be like, okay, what can we do now? Right? And that's where we start really, because, you know, AI is great at those repetitive tasks. And as humans, we are too focused on a lot of low cognitive tasks. And it's very sad because we as humans, we are problem solvers. The AI is a great context and like it has so much context of the world, but it doesn't have your context, but you have your context. So you gotta find that middle ground. And that's basically, Working with AI to reduce or hopefully even remove those repetitive tasks so that you can start focusing on the higher cognitive work. And as you focus on the higher cognitive work, first of all, I think that if our brains focus on the higher cognitive work, we get smarter as we go. And this is where AI can't dumb us down, right? Because everybody talks about how AI is going to dumb us down and so on. But if we let the dumb stuff AI do, and then we focus on the higher stuff. I think it's a win-win.

Yeah, but it's just, uh, firstly beginning with a little bit of a brain dump, and then once you get some of those ideas out, what do you find is a great way to prioritize those and think, what are we going to tackle first, and how do we go about solving some of these problems?

I think that's when we can tap back in the old frameworks like ICE, Impact, Confidence, Ease. These are experimentation frameworks. They're prioritization frameworks, right? They're scoring your experiments. I used to be a growth consultant before AI. So, you know, when I was in growth consultancy, the idea was get all these ideas and then score, prioritize the ideas. And you have frameworks, you have RICE, ICE, PIE, Probability, Impact, Ease, you know, and you give it a score. And then as you score every idea, well, score it with your team, not just yourself, right? Let everybody give a score, then do an average of those scores, and you will find the top 3 things that you need to work on, right? And usually they're the most impactful, you're the most confident about this happening, right? And they're the easiest ones.

Yeah, I love that. And then the different ways about solving some of these, obviously quite a few of those methods you mentioned focus in first around ease and making it, you know, simple, like you want high impact and hopefully something that's easy to get started with, especially with AI. It can be a little bit of a stretch. There is some easy solutions people are familiar with, things that are already in their Office suite, whether that's Copilot or Gemini or something like that. There's some low-hanging fruit there. But how do you think about kind of like the the scale when it comes to ease of rolling these out?

Yeah, well, I think, you know, when we say like Copilot, or I'm going to say potentially more about Claude today because Claude has something called Claude Cowork and can potentially do about 70-80% of the things that we do on a browser, right? So a lot of those tasks that we're talking about, like reporting and going and finding information on an analytics dashboard, taking that, putting in a spreadsheet, then making a report out of that, that can already be automatable, right? And a lot of people are like, oh yeah, but I don't trust AI, you know? And it's like, well, AI now does better work than we do. Humans make a lot of errors, you know, and nobody thinks about humans making errors, but they're like, AI makes errors. By the way, AI doesn't make errors, there's hallucinations, right? And hallucination is a feature, it's not a bug, right? But AI can write code, can write a script, a Python script, which is a programming language for data analysis. Like it's a great one for data analysis. And then it can run that script and that script is deterministic. So you're gonna get real results. You're not gonna get, you know, hallucinations. You're not gonna get wrong, false information. So there's a lot of easy, easy, easy tasks that we don't think, and we all wanna really think I want agents that is going to replace my whole SDR team. Well, hold on, hold on. Don't replace your whole SDR team first. Please don't do that, right? Because after that you're going to be sad. Instead, look at what your SDR teams do that are painful, that sucks, that are boring, right? And that would be high impact. Why high impact? Because they wouldn't have to do that so they can do more sales, they can do more reach out maybe. So when you really prioritize those and you'll see that there's so many easy things, but it's just that we're so used to doing that ourselves that we can't really see AI doing it for us or with us. And this happens even to me, but in a different scale, of course, you know, it's like I look at certain things and I'm like, I catch myself, I'm like, why am I doing this? And then I zoom out for a second and I'm like, okay, well, maybe ChatGPT can't do that, but I know that Claude Core can do it. I know that Claude Code is on the terminal, it's a bit more scarier. It says code, but it's not for coding only. I know that that for sure can do it. And then I remind myself, and then I mix these AI tools, and I mix them, and I work with 2 or 3 of them, and I get the task done. And potentially that task that took me 5 hours, but now it took me 30 minutes.

I think one bridge that people need to really cross to get the most out of the AI. And I'd love your thoughts on it, is to go from initially you start using it and you think, oh, I want to do this thing, and you ask AI to do it and it goes off and does the thing. And then you think, oh, I want it to do this next thing, and then you ask it to go do that, and then it goes off and does that. And it's much like when people start building out their first few team members and they're just giving them a task, they do the task and then they come back. And there's a shift that happens as the business owner starts to grow where he starts to think in terms of responsibilities and then he's now assigning that to someone and they're doing those things on a repeated basis. How do you think about that with AI? Because that's ultimately where we want people to get to is is setting up recurring tasks. Like, the AI doesn't have to be told to do it, it just does it on a set rhythm.

I think that first of all, I would like to say that it is tough, and you know, I accept that because we're redesigning how we work, right? We have to rethink about everything that we do day in, day out. If we do, especially, we're going to have a great time because we have this incredible technology with us. So this is why as a business we're even removing the word using AI and we're focusing on working with AI. Because when we use AI, we just give it a task and we want it to be done and then we're not very happy with it. But when we work with it and we give it the right context of where we're at, or as we were just discussing earlier as well, you know, you said knowledge base, you know, what is your knowledge base? What's your source of truth? You know, when I speak to businesses a bit bigger, 450 people, they have about hundreds of tools. You know, they'll have even 2, 3 task management tools and the context is all over the place. So then the AI can't really work with you to find those things. So first you have to really clean house, you know, and do that. But also, as you mentioned, I think that, you know, job descriptions are a real thing again, you know, and I think you got to have a job description and you can go very specific about these jobs now. So when we talk about agents and we're like, oh well, I want a marketing manager agent, that's wrong. But if you want a LinkedIn A webinar planner agent. That's correct. So how I think about these, I'm going to go a little bit in my field, old school, which is a year ago when custom GPTs were cool in ChatGPT, you would build, I say old school because for us it's very old school now. You know, it's like, wow, when was that? That's like long, a long time ago. It's not right. It just moves so fast. But custom GPTs, I would always say, all right, let's say you're a solopreneur. You're a one person band, right? What's the first 10 people you'd hire? You'll tell me somebody in sales, somebody in marketing, somebody HR maybe, right? Because you want to scale somebody in operations, maybe, maybe a customer support if you have something like that, right? So you plan your first 10. That's fun. You know, you do that and then create custom GPTs or agents, whatever in the beginning, or maybe even ChatGPT threads or Claude threads, and then work strategically. And then you're going to have maybe inside the marketing manager, you're going to have about 10 recurring tasks, right? Every week you do a webinar, you have to do some social media, you have to do some ad campaigns. What else, right? You have a newsletter. Here we go. We got already 4, right? And I can map this out. You map these activities, right? The jobs to be done. You map these out and now you can design custom GPTs or agents or threads for those things. And then maybe you have now a billion dollars and you can hire 1,000 people, right? Then you're going to want somebody who does just design work for the webinars. You're going to want somebody who just plans or just writes, uh, you know, a piece of article every day. Somebody that announces the news to your clients or something like that, right? You can go niche or niche or niche, and then you design an agent or a thread for that. So you can go deeper and deeper because the deeper and deeper, the niche you get The AI does a better job, right? The narrower the task is at hand, the better context you can give about that narrow task, a clear process, SOP, standard operating procedure about that task at hand, the better the AI will be. And it is difficult because where our brain is getting fried, especially mine right now, is suddenly I'm managing hundreds of these. I have 146 agents doing work. and they can all ping me when they need me, right? And so now, but, and there's no middle management layer anymore, right? It's me working with 146 agents, right? And not humans, but agents, and they're all like, they need things from me, right? And then as you do that, let's say that you run that 10 times, 15 times, 20 times, you can start moving into more automated ways and say, do this every Friday at 3 PM. And why after 10 times? Because the first 10 times The quality is going to be very low. Like the 3 times is going to be very low. The 4th, 5th, 6th, you're like, oh, okay, it's getting better because you have to keep fixing that job description, fixing that SOP, giving it more context. Oh, it made a mistake. Why did it make a mistake? Because it doesn't know. It's a child. It wants your training. It wants you to tell it exactly how to do that. And the more you give that love in the beginning until that 10 times, 15 times, maybe sometimes if it's a very important kind of SOP and a bit more complex, it could be 20 times. But at some point you're like, oh wow, that's really good quality now. And then you see that quality over and over again and you're like, you know what, I can let this run wild now, right? And then you schedule it and then you let it run wild and then you're very impressed, you know, uh, and then you go on to the next one. And as humans, it's very difficult for us because we're used to You know, letting humans kind of figure things out, which is great. AI can figure things out, but with your supervision, with your context, with you really tweaking those things, it's a beast. It's an incredible thing because it goes from that, that being that baby, that child that is, you know, 3 or 4 years old, suddenly goes to being a teenager and very quickly to a university degree person. Right? And this doesn't take 18 years, it doesn't take 18 months, it potentially takes 18 days to get there, you know, if you give that time. So we gotta give that time today if we wanna save the time tomorrow.

Ah, hey, I know you are tuned in right now, so I'll keep this super quick, but you like awesome YouTube videos, don't you? Well, there's not really that many YouTube channels that create great content about how to systemize your small business. So I want to make sure that you catch every video that I publish. So if you haven't already, firstly, give this video the thumbs up and then go ahead and subscribe to my channel. You can even hit the bell notification to make sure that you're the first to know when I publish anything. So go ahead, do it right now. I'll wait a sec and then I'll transfer you back to the video. I love the framework that you've walked us through there as far as starting off with the pain points and then going to work on those, effectively building out your SOP or figuring out what it is that you want solved, running with the AI, letting it do that, iterating, getting it closer and closer. And then automating, and then you can start to see how this would build on top of itself, uh, and different agents would feed into each other. You've been doing this for quite some time, and like you said, it's just moving so quickly. Um, is there anything that you would do differently, um, or the way that you would approach AI now with what it is that you've learned? I mean, you've given that framework, so I'm assuming that's the framework which you approach it, but is there anything else that you would have done differently as you were learning?

Well, I mean, I think that, you know, if I said no, it would be potentially a lie, but what would I have done differently? It's difficult because it changes so much that everything is different all the time, right? So when I look at when I started the business, which is not long ago, it's what, a year and a half, close to 2 years now, it was just a year after ChatGPT or something like that came and it's gone so fast. We've helped so many clients so rapidly and what I said to a client a month ago and what I would tell them today is already different. So I don't know what I would have done differently, but I think that from sooner I would have made mistakes, right? And I think back in the day, back in the day again, you know, I was like expecting a bit more perfection and I was like, oh, okay, 'Well, it can't do this. Let me—' Like what I'm saying now to people not to do, and it's like, but let— do that mistake, but do it rapidly. Make sure to look at it really well and understand why it did that mistake and then iterate on it. I think in the beginning I was more like expect— I was waiting for better models or I was waiting for better tools, but it was already doing some simple stuff really well. And then I was thinking that it would do more stuff well, but we all are expecting much more from AI. So I think that's what I would have done actually. I would have not expected that much more from AI. I would have been like, okay, it's great at doing this easy task, which is labeling my emails. Yeah. Let it label your emails. Let it draft some emails for the simple ones. I didn't do that. I was like, well, I do my emails better, you know? And a lot of people tell me, I do this better. I'm like, well, good luck. Well done. Done. You think you do that better, but we can tweak it 10 times, it's going to do it better than you. I mean, the world's best developer said that he's scared because it codes better than him. I said, oh my God, I can't code this fast, this well. So imagine if the world's best developers are saying this, your financial task of looking at last month's revenue or your task of answering a couple of proposals or agreements or reading them, do you think it's more complicated than the world's best developers' work? I don't think so. I think that I had a bit of ego as well, maybe, and I'm starting to lose that ego because these machines are very intelligent, again in air quotes, because it's artificial intelligence and it knows a lot, but it doesn't know what to do without you. Yeah, right. So I don't need to be that super intelligent because our brains anyways is not meant to store information, it's to solve problems. But this artificial intelligence thing, it's to store a lot of information, to be able to gather that information, to be able to generate answers from that information, uh, quite quickly, quite rapidly. So when we start understanding the differentiation of us as humans and this machine and this technology or whatever that is and it's going to be. When we start understanding that, we're going to have a better relationship, you know. And I think that's how I was thinking a couple years ago to what I'm thinking today changed a lot. I wish I knew what I know today back then, but I couldn't have because, you know, I had to get in the trenches and learn and make those mistakes. But I just wish that I've done them earlier.

Yeah, yeah, I think that's, uh, sound advice. I feel like, um, A lot of people are like just tentative and stepping back when they really just need to jump in and play around with it, get comfortable, see what it can do. And very similar to when sometimes you're either hiring a new team member or trying to get someone started, oftentimes people look for this unicorn person that can do everything and do it just right. And I think they'll oftentimes think about outsourcing or giving that team member the hard thing to get it off their plate, when you're much better off giving it all the easy things first to free you up to work on the hard things. It's a much better way to think about it.

Yeah, I think, I, I think you nailed it on the head because it's, you know, when we hire someone, somebody junior, and we're like, oh well, I've hired this person, but I can do this better myself, so let me do it. Well, no. How's that person going to learn that task? They're going to do it, and you gotta supervise, and you gotta help them, and you gotta help them optimize that task. It's the same thing. It's— we're doing the same thing again, um, and we just gotta accept it. That it's— the news make it sound like it's like, wow, these machines are incredible. They are incredible, but your first 10 times, 20 times that you interact with it are not going to be incredible. And then you're gonna kind of all give up or you're gonna get frustrated. Um, but if you, if you keep, if you keep doing that, that task or whatever that you're doing, it gets better and better and better. And then hopefully you're gonna have your wow moments. You know, when we do our training, our behavior change programs, my number one focus is to make sure that everybody has a wow moment, an aha moment. And you know, it's my biggest job actually, it's to inspire people. At work and I inspire them. They go, oh wow, I did this and it was incredible. I've never been able to do this before. Done. Now I know that they're gonna work with AI, they're gonna try new tools, they're curious, right? So sparking that curiosity is actually my main task and I think that's what the clients pay us for. It's like Tim's gonna come in and he's gonna inspire everybody and everybody's gonna be suddenly super inspired about working with this technology. And then when I leave, companies still, and their team members still innovate a lot because they have this curiosity now, and they know that this is an incredible time to be alive, and that the news and this negativity is just a lot of media. And yes, okay, I will accept there might be negative outcomes at some point because, not because of the AI, by the way, because of humans, how they're going to use it. Right. This is the same thing as guns in the US, right? Like, you can use it for bad things, right? It's a bad thing. You know, AI is so much more positive if used well, if used to reduce those things, to help us. I think that, I mean, I'm so excited to be alive today at this time and age and be at the forefront of this because I think it's going to be a huge transformation for all of us. It's already happened. I see it and I get tunnel vision, you know, I get, you know, I think that on LinkedIn, you know, it's an echo chamber, you know, where 10,000 people who talk about AI all the time, I guess. And I think that everybody knows about AI. And then I get a new lead and then we start a new program and I meet hundreds of new team members, right? And I survey them, we send the survey and then I see some people say 1 out of 10 on their comfort of working with AI. Some say, oh, they use it. 'Once a month,' and they did some emails. I'm like, 'What?' And like, and then I— then it's like reality call, you know? It's like, 'Tim, you know, you are in this, you know, in this rabbit hole, in this, in this tunnel. You think everybody's doing AI, but no, the world is still not there.' It's just the news as well, and what's happening around us just talks about this a lot, that we do believe, um, we're late, and that we have all FOMO, uh, fear of missing out. And, you know, I always show this, it's a love, not fear kind of coaster, you know, and we have this movement is every decision we make out in life is out of love or out of fear, right? And if we make this decision out of fear, we are not going to have a great time with AI. But if we make it out of love, there's a very big chance that we're going to come out on the positive side as humanity, right? So I believe that more good people need to work with AI so that the bad people that are going to work with it are not as many as the good people working with it. You know, I think it's definitely a very big political subject, as we know, and it's going to be big divisions. But, you know, you've got to choose on what side you want to be on.

I think the good news is you are one of the good guys. So I appreciate your time today and sharing. I'll put the link through to AI Operator. Is that the best place for people to go and learn more about your work?

Yeah, arapreda.com, on YouTube, Tim Kakir on YouTube, on LinkedIn. I'm very vocal. I have a newsletter for free every week. You know, I, as you said, I think that, you know, I made the mission in my life now. Finally, I have a mission in life and is helping humans with, especially professionals at the moment. I have other ambitions after this company, but right now is helping professionals have a better time and have a better day and have better work thanks to working with AI, you know. So please do follow. And I answer every question, every comment, even if it takes time, I do answer them, not AI, by the way. I like to still, I can still manage it, you know. So please, yeah, do reach out if you need any help because I know it's tough what we see out there. It's scary, but it doesn't have to be. Yeah, perfect. Thank you, Tim. Thank you. Thank you very much for having me.