Ep 32: The Transformational Impact of AI Agents and Agentic Workflows with Tyler Fisk

[00:00:00] Welcome to Elevate Your AIQ, the podcast focused on the AI-powered yet human-centric future of work.

[00:00:04] Are you and your organization prepared? If not, let's get there together. The show is open to

[00:00:09] sponsorships from forward-thinking brands who are fellow advocates for responsible AI literacy

[00:00:13] and AI skills development to help ensure no individuals or organizations are left behind.

[00:00:18] I also facilitate expert panels, interviews, and offer advisory services to help shape

[00:00:23] your responsible AI journey. Go to ElevateYourAIQ.com to find out more.

[00:00:39] Hey everyone, it's Bob. If you've been listening to the show for a while, thank you very much. I really

[00:00:43] appreciate it. Feel free to drop me a line if you have any feedback at all. If this is your first time,

[00:00:48] welcome to the show. Today's episode is with Tyler Fisk, who's the co-founder of not one but two

[00:00:53] AI consultancies, Light Magic AI and Galactic Ranch, where they design, build, and implement

[00:00:58] AI solutions for companies of all sizes. Tyler has always had an entrepreneurial spirit,

[00:01:03] whether that started with farming, believe it or not, before switching gears to build

[00:01:06] e-commerce businesses and later realizing the promise of AI. Tyler is designing some really

[00:01:11] innovative solutions that show what's possible with AI and agentic workflows, which is a hot topic,

[00:01:17] making AI accessible and practical for his clients. I always love chatting with and learning new things

[00:01:22] from Tyler, so let's get to it. Thanks for tuning in. Hey everyone, it's Bob Pulver. Welcome to another

[00:01:28] episode of Elevate Your AIQ. Today, I am joined by my friend Tyler Fisk. How you doing, Tyler?

[00:01:34] I'm doing awesome, man. I'm here with you. Thank you for having me.

[00:01:38] Absolutely. It's great to have you. You're down in the Carolinas?

[00:01:41] I'm actually in Middle Tennessee. I'm in Cookville.

[00:01:44] Okay. So were you spared from the storm?

[00:01:48] For the most part, yeah. I had some friends that are over in western North Carolina that got hit really

[00:01:53] hard. Where I live is actually kind of up on a mountain, so we didn't really have any flooding

[00:01:58] issues, but the wind got really wild here for a little bit. So we had some trees come down, but

[00:02:04] that's nothing compared to what I've seen on the internet.

[00:02:06] Yeah, no, it's pretty bad. I have a friend in Charleston who also has a house in Nashville,

[00:02:11] and he's actually not sure what's going on with his property. So hopefully they'll get communication

[00:02:18] restored and they can figure that out. But fingers crossed that goes smoothly.

[00:02:24] Exactly. So I know you're involved in a couple things related to AI development, upskilling others.

[00:02:33] You're spreading yourself around a lot of interesting projects. So why don't we kick

[00:02:38] things off with just you giving a little bit about your background and some of the entities that you're

[00:02:44] involved with. Sure. Yeah. So, well, my previous 20-something year career was mainly in the tech space,

[00:02:52] but I always joke saying that I didn't really work in Silicon Valley, even though I worked with a lot of

[00:02:57] folks from there. I worked in the Tennessee Valley. So it was an e-commerce background mainly across a bunch of

[00:03:04] different brands. So my family has a greenhouse manufacturing and gardening supply company.

[00:03:11] And my dad started that in our spare bedroom with myself and my brother and my then girlfriend,

[00:03:18] now wife with just 72 bucks. It was very, very humble beginnings. And now that is one of the bigger

[00:03:25] names in the space and it's an eight figure like ARR type company. And I tell that story a lot to kind of

[00:03:32] set the stage that I got to go through an education process of what is it like to run a tech first, internet

[00:03:40] first business and to wear all of the hats to make all that stuff happen. And also in the later part of

[00:03:47] that career, I was able to start dabbling in AI products really before the chat GPT boom. I was using

[00:03:54] Grammarly a lot in its early days. We had some products that we used with some of our Shopify stores

[00:04:00] that were using some machine learning algorithms. And also I think he was even using GPT-3 like an early

[00:04:06] version of that to be able to write email copy to our customers that sounded like us. Basically,

[00:04:14] it was it was decent. It was more templatized than it was completely generative, but it was still pretty

[00:04:19] solid. And that lit a fire in me to learn more about AI. So now fast forward, you're correct. Like

[00:04:27] I'm in a couple different things. I'm an entrepreneur at heart. And so I'm a part of Light Magic. Light

[00:04:35] six of us in total that we all met in the AI exchange. Shout out Rachel Woods and the AI

[00:04:41] exchange. I believe that's even where we met, isn't it? That's what I was thinking.

[00:04:44] I think it might've been. Yeah.

[00:04:46] Yeah. And so that is a really cool community online. And so the six of us met there,

[00:04:54] we started as an accountability group, and then we turned that into what would it be like to work

[00:04:59] together. And so we've done several projects together. And since then, our personal brands

[00:05:04] have taken off and our personal lives have gotten really busy. I've got two young kids and a few of

[00:05:11] the others in the group have kids as well. So we have have really had to focus more on our personal

[00:05:17] brands and we leave the Light Magic brand mainly for big projects that come through like more

[00:05:22] enterprise level or things that just ourselves and our teams can't do individually. So Galactic Ranch

[00:05:29] is my personal brand. And that is where I do AI development, AI consulting, even partnered up with

[00:05:38] Sara Davison and Hunter Canning who are also from Light Magic. And we've got this new Maven course.

[00:05:44] This is our first time doing teaching. We're in the middle of cohort one right now. It's been a lot of

[00:05:48] fun. And it's called how to scale your business with AI workflows. And so that's, that's, I've got

[00:05:54] my, I've got my hands or fingers in a lot of pies, I guess is the best way to put it. So we like to stay

[00:06:00] busy. Oh my God, that's awesome. No, I love hearing the sort of origin story and people that didn't

[00:06:06] necessarily have a tech background, but recognizing the opportunity and recognizing a way to either bolster

[00:06:14] your existing business and give that some competitive advantage in a sort of digital and technical edge,

[00:06:22] if you will, but also exploring other more deeper tech kinds of solutions. And also, you know,

[00:06:30] just upskilling, obviously the theme of this show is really how do we elevate people in all walks of life

[00:06:36] and get them to recognize some of these opportunities. So I think you're setting a good example and a

[00:06:43] relatively high, high bar. Thank you. Yeah. So, you know, you mentioned you're using AI before,

[00:06:50] you know, ChatGPT entered the public consciousness. And that's one of the elements that I talk about

[00:06:58] when I try to educate people. So when you start to teach people, maybe it's too early, but when you get

[00:07:07] feedback on either the Maven course or, you know, earlier, you know, more informal, you know, training,

[00:07:14] are people getting it? Are people anxious to tinker like you did when you first started? I mean,

[00:07:21] what's the audience sort of reception and their level of adoption, I guess, before they encountered your

[00:07:29] courses?

[00:07:30] For previous teachings that we've done, and even just some of the seminars and things that I've done,

[00:07:37] always kind of start off and set the stage of saying, I'm about to tell you a lot that's going

[00:07:44] to really make your mind melt. And it's, you're going to be overwhelmed, I can guarantee you,

[00:07:48] I still feel overwhelmed. And I do this every single day. So I always kind of set that because that feeling

[00:07:53] never really goes away, at least for me, it doesn't. It really speaks to the pace of how fast,

[00:07:59] this stuff is moving. It's, you know, I've been in tech my entire life. And I'm used to this, like,

[00:08:04] this cycle of the new thing coming out. And anymore, that's not the case. But I really have

[00:08:10] been trying to focus my education and even more recently, a lot of my implementation work on small

[00:08:16] and medium sized businesses, for a couple of different reasons. And that's because so many people

[00:08:21] I know are interested, and they know how I've got to do AI, I've got to learn AI. And that's literally

[00:08:26] how they'll describe it, or they'll call it chat GPT. And those are kind of the, the nouns that

[00:08:31] they're throwing around right now. And just generally kind of like what you were talking

[00:08:34] about with the jargon or whatnot, which is totally cool. And that's, that's where they are in their

[00:08:39] journey. What we've found is that people definitely have a hunger and a desire to learn. And at the same

[00:08:47] time, they don't fully know what's possible or what the state of the union of AI really is now.

[00:08:53] Because in those same seminars, I'll start off very similarly to like what you were just talking

[00:08:58] about of the background. I'll even go all the way back to like Alan Turing, you know, 1950s,

[00:09:03] work all the way up and kind of give this background just to be like, this is not something new.

[00:09:08] We are literally building on the backs of people that have come years and years,

[00:09:14] dedicating their whole lives to this, to get to the point that we are now. And then we've hit this

[00:09:18] instant boom and open AI is really good marketing scheme here. And that's where it's gotten y'all's

[00:09:26] attention. So it's, it is a very powerful tool. People are really wanting to learn it. They don't

[00:09:33] know where to start. They do feel overwhelmed. And if, if they have much experience with it at all,

[00:09:38] or what they would call it experience, they've chatted with chat GPT a bit. So saying that once you

[00:09:44] hit that point, then you kind of hit like this diverging path of folks that want to learn about

[00:09:49] it. So they have an understanding of it, but they don't want to do it. They want to either like empower

[00:09:55] people on their team to be the AI, AI folks on their internal team, or they're wanting to hire people

[00:10:01] for it. Then you're also going to get a lot of the folks that are like us that are, you know,

[00:10:05] tech nerds and AI nerds, and they're going to go all the way down the rabbit hole. And they want to

[00:10:09] really learn this stuff inside and out. But that's, that's kind of where I see it. And that's

[00:10:14] the way it was even in the AI exchange. When we had folks coming in there, it was some of them were

[00:10:19] coming just to really get a better grasp and understanding of what this is and how to use it

[00:10:24] and what the terminology is. And then also a lot of them are really taking the time to learn prompt

[00:10:30] engineering, automation, workflow engineering, and even coding. A lot of folks are learning how to get

[00:10:36] into like Python and stuff now, just because they recognize that as the, the, the default language

[00:10:43] that is, that works really well alongside with, with a lot of these LLMs. There's always going to be,

[00:10:48] you know, I'm not a learning and development professional by any stretch, but, you know,

[00:10:52] people have different approaches to learning. They learn and through different means, right? They learn by

[00:10:58] doing, they learn through, you know, online courses, instructor led or self-guided and things like that.

[00:11:04] And so, so you've got that dynamic still at play, I think. And then obviously you've got some people

[00:11:10] with a, with a learning mindset and people who are anxious to make sure that their, their own career,

[00:11:17] you know, trajectory and potential is not, you know, disrupted by, by technology. And they recognize that,

[00:11:25] you know, this is the future of, of work. I mean, not every role will be impacted at the same speed or to the

[00:11:32] same degree, but it's not something that's going to go away. I mean, the hype is there and for many

[00:11:40] situations it's legitimate. I mean, it's not a, it's not a trend. I mean, it is a trend, but it's

[00:11:47] not something that's going to just, you know, dip and, and go away. So I think we do need to embrace it.

[00:11:54] I think part of where people get confused is because there's so much to potentially learn.

[00:11:59] And they've got to at least figure out what are the specifics of what I need to know. Like I kind

[00:12:06] of use the, the automotive analogy, like I need you to get your permit and your license before you

[00:12:13] drive. And you should probably learn how to change a tire, know what some of the warning lights are or

[00:12:19] whatever, but I'm not asking you to be a mechanic. Right. So I think you want to, you want to be self

[00:12:26] sufficient to a point, right? You don't necessarily need to completely pivot and become,

[00:12:31] you know, an AI guru, but it can certainly help you sort of hedge your bets as your career may take

[00:12:38] some unexpected turns and hit some bumps. That is something I've seen a lot over the last year.

[00:12:45] I've made it part of my weekly routine to try and have zoom coffees with different folks that I've met

[00:12:50] on the internet. And a lot of those folks have been downsized or been given packages to, to retire

[00:12:59] early, all that sort of stuff. And part of that is because of just the current economic landscape.

[00:13:05] Part of that is because these companies aren't saying the quiet part out loud, they're actually

[00:13:09] rolling out AI and they're replacing people with it. And when we're doing our implementation and work,

[00:13:17] that's a vibe check that we really do on our first discovery call with folks that, you know,

[00:13:23] if you're going to take these new superpowers and you're going to send that straight to the bottom line,

[00:13:26] and you're going to just cut your workforce in half, instead of trying to help them, help them learn,

[00:13:32] help empower your people so that literally everyone wins at the end of this, then we're,

[00:13:37] we're not the team for you. There's 100%. There's plenty of teams out there that will,

[00:13:41] that will go and do that. They, they won't think twice about it, but for us, it's, it's just an

[00:13:47] ethical thing for me. And by no stretch of the imagination, do I think that every business that

[00:13:53] we work with at some point is not going to lay someone off for one reason or another, but we

[00:13:57] just want to do everything in the realm of possibilities of the way that we can affect

[00:14:01] these relationships is that we inspire them to have the conversations and build up the culture within

[00:14:06] their team because folks are afraid of it. That's, that is the number one thing that I've seen.

[00:14:11] It's, it's almost, we could turn it into a drinking game. When I tell people what I do for a living over

[00:14:16] the last couple of years, uh, they'll mention Skynet within the first five sentences. It's like almost

[00:14:21] always that comes up. And, uh, at some point they're going to, in one way or another, ask me if

[00:14:28] they think that AI is going to be able to replace their job. Long story short is yes, it probably can,

[00:14:34] but at the same time, part of my job is that we try and figure out what's your secret sauce in your

[00:14:41] organization that the people do the best, or what is the thing that y'all really love doing and you

[00:14:46] want to do more of, uh, that is what we can help you figure out how to kind of build this harmony up

[00:14:52] between you and the AI systems or agents or AI team members is even how we're calling them now. And we

[00:14:57] stand up some of these agents so that y'all work in collaboration with each other instead of

[00:15:02] being fearful of if they're going to come and take your paycheck and then you're going to be out on

[00:15:07] the street because that, that is a real concern in organizations that we've been in for sure.

[00:15:12] Yeah. I think in a lot of cases people, and I'm not saying that's not a legitimate

[00:15:18] concern. I think like a lot of issues, people see that this could be a slippery slope, right? Like, well,

[00:15:25] if it starts taking on, you know, these couple tasks and then at the pace in which it's gaining

[00:15:33] momentum and, and gaining other capabilities, like as we've seen with open AI now has chat GPT now has

[00:15:41] some better reasoning capabilities with the O1 release. People see you're basically agreeing with

[00:15:50] me that this is the tip of the iceberg, right? So if it can start to do this and it can start to

[00:15:56] narrow down these choices and, and be this effective in decision-making, not just generating content,

[00:16:02] but actually doing design work, you know, writing music, making images, and now the multimodal

[00:16:10] capabilities where it can do some of these things, you know, simultaneously, it's starting to creep up on

[00:16:17] things that we thought were clearly within the domain of, of a human, you know, brain and human

[00:16:23] capability. And so I totally understand why people are nervous to, you know, today it's automation

[00:16:31] taking, you know, these three tasks off my plate tomorrow. It's leaving me with only 25% of what my

[00:16:39] job was. And then the company has to take a look and say, what are we doing here? How are we going

[00:16:45] to redesign work? Are we thinking about the humans, not just like humans in the loop, but the human

[00:16:52] centric nature of work and what are the implications to employee engagement? What are the implications to,

[00:17:01] you know, how we contribute to our, you know, our community and, and, you know, the world at large

[00:17:07] with our products and services? What are the consumer, you know, client expectations for working with actual

[00:17:14] people or just trying to get their stuff done, however it needs to get done? So there's a lot of

[00:17:20] sort of existential kinds of questions that people are actually, you know, sort of bringing up. So,

[00:17:26] you know, I don't have all the answers. I do think, you know, the Skynet thing, it's kind of funny that

[00:17:31] people still, you know, think about that because I still think that's, that's pretty far away, but

[00:17:39] I don't know. I just think the more educated people are about what's, what's actually happening,

[00:17:43] the more unrealistic that seems, but yeah, I mean, I guess if, if you're doing some long-term

[00:17:50] thinking, you know, maybe Skynet won't affect us, but maybe our kids, I don't know. No, that's,

[00:17:57] that's fair. That's what, you know, something that we've, I think we've talked about in the past and

[00:18:01] it's something I talk about with folks in the space a lot is that my boys are just turned three

[00:18:06] and they're, my oldest is about to turn five. And the world that they are going to grow up in

[00:18:11] is just completely going to be different. Like it unimaginably different. I can't do a five-year

[00:18:18] plan. I don't feel like I can accurately do a five-year plan anymore because that's how fast

[00:18:22] the world is moving. It feels like at this point on a bunch of different fronts, not just because

[00:18:27] of AI, but definitely AI has something to do with that. And so one of my missions that I've kind of

[00:18:34] put upon myself is that I want to get as many people involved in this space, or at least aware of it so

[00:18:40] that they can bring their voice to the table for a bunch of different reasons to, because collectively

[00:18:45] we're the ones that need to work through, how do we deploy this stuff? What's our culture around this?

[00:18:51] Like all these big existential questions, how do we actually work with AI responsibly?

[00:18:57] And I know you've probably seen here in the news, even in the last like week or so,

[00:19:02] open AI's the rest of the, well, almost the rest of their, their initial founding team has left.

[00:19:10] They're making the switch from a nonprofit to a for-profit company. What's the implications of

[00:19:16] that going to be? Because now they're going to have, you know, shareholders that they need to,

[00:19:21] it's a whole different ball game on the level of responsibility that they're going to have on that.

[00:19:25] Same thing with the California law that just got up and got vetoed by Gavin Newsom.

[00:19:32] I haven't fully made up my mind on how I feel about that yet, but essentially it was going to

[00:19:38] put some guardrails around how California deals with all these AI systems and the startups and how

[00:19:45] it can be rolled out. And Gavin Newsom vetoed that because he was afraid it would stifle innovation.

[00:19:50] And also, by the way, I can't remember the exact number. It's like close to 15 of the top 30 AI

[00:19:58] companies in the world are centered in California. So that might have a little something to do with

[00:20:04] that ruling there, but that's like very in-depth industry stuff. And that's very focused in closed

[00:20:11] rooms in these research labs or these organizations that are developing these things.

[00:20:16] And I personally think that this needs to be a more inclusive and democratic process,

[00:20:22] not only just because I think everyone can benefit from learning and using these tools in a big, big way,

[00:20:28] but they're not, like you said, they're not going away. And it's, they're going to be a big part of

[00:20:33] our future and our children's future. And if we decide that we're just going to stick our head in

[00:20:39] the sand, and if we're going to be like, well, I'm going to get to that someday at the pace that this

[00:20:44] stuff is moving, that someday is going to get here quicker than you realize. And maybe not,

[00:20:50] this is not like a fear thing, but you're going to see more and more people adopt it. So if you don't

[00:20:54] get on the learning train to really start at least get a basic level understanding of it quicker,

[00:21:01] I hope that it won't be that folks can't catch up, but that, that is a real concern of mine.

[00:21:05] Or even more so, if you go back to when we were younger in the earlier internet days,

[00:21:11] there were a lot of companies out there and a lot of people playing in that space. And now you don't

[00:21:16] see Ask Jeebs anymore. Yahoo is not the superpower that it used to be. MySpace is not the superpower it

[00:21:22] used to be. All of that power got consolidated down into the fang stocks, basically. And I don't want

[00:21:30] to see that happen in the AI space because now we're playing with super intelligence. So what

[00:21:35] happens when super intelligence gets consolidated to the hands of a few? I don't particularly like

[00:21:41] that future. Yeah, no, you're bringing up some really important points. I think the responsible AI piece

[00:21:47] is critically important. I think, you know, that's one of my focus areas. So, I mean, just if we think

[00:21:55] about those fears that you mentioned before with, you know, Skynet and things like that, responsible AI

[00:22:01] practices, governance, the appropriate, you know, legislation that's, you know, hopefully consistent,

[00:22:10] more consistent than the patchwork, you know, we see today. But we need all companies developing

[00:22:17] AI and all people who are building with some of these AI solutions, which could be any of our listeners at

[00:22:25] this moment, right? You could be whipping up a custom GPT, you know, an agent, maybe you're on AWS,

[00:22:31] you know, Party Rock, just, you know, having fun, but you're creating something from scratch where,

[00:22:38] you know, where are you getting the data that you're inputting? Is it copyrighted? Is it proprietary? I mean,

[00:22:44] you know, this is part of the education, right? Part of the upskilling is not just what it can do and how to

[00:22:50] better, you know, prompt it or whatever. It's not just about your AI skills, it's your AI literacy and

[00:22:56] being responsible by design. And what does that mean? So that we're making fair, you know, ethical

[00:23:02] decisions based on the data and algorithms that fed these AI tools. So all these things are going to

[00:23:08] mitigate, you know, risk in general, and in particular, some of the bigger, scarier things

[00:23:14] like you're talking about, like moving towards artificial general intelligence or artificial

[00:23:19] super intelligence, whatever it is, we've got to stay human centric as we do this.

[00:23:26] The California thing, I didn't read through the entire original bill as written, but I understood

[00:23:33] the premise. And I did read Gavin Newsom's letter of why he was vetoing it. And I thought it was very

[00:23:40] well thought out and appropriate. The problem was it was putting all the onus on the biggest

[00:23:49] large language models with the most compute power and, and what have you. And he, I thought rightly

[00:23:57] called out the fact that who's to say that tomorrow someone doesn't build an incredibly powerful,

[00:24:04] you know, large language model that, that uses a fraction of that compute, or what about the high risk

[00:24:11] use cases, as opposed to like you built that computer too big, you know, you're going to get

[00:24:16] regulated. But these guys who, you know, networked a bunch of smaller ones together can completely

[00:24:22] circumvent the legislation or someone could use a small language model or some other approach and

[00:24:29] still come up with something that is like basically open source and, or, you know, on, you know, 4chan or

[00:24:38] some back channel where people can take it and apply it to do, you know, facial recognition

[00:24:45] on closed circuit TV at stadiums or airports or like, you just don't know. So it left too much

[00:24:52] to chance and gave the public a false sense of security. So I just predict there's going to be

[00:24:59] people who are going to take that knee jerk reaction to the headline that says, I can't

[00:25:03] believe Governor Newsom is, you know, putting us all at risk. But if you look at it in detail,

[00:25:11] I thought it was a very thoughtful and measured and logical response. He wants the legislation,

[00:25:18] but it's got to be more buttoned up. And it's got to think about high risk use cases,

[00:25:22] which is the way that eat the EU thought about the risk level in their sort of risk pyramid for

[00:25:28] the EU AI act. What are the actual use cases that are going to be used and what are,

[00:25:33] and how do we build upon like GDPR and protecting consumer rights and personal data and worker rights

[00:25:40] and things like that. So it's a good talking point, but we've got to.

[00:25:45] Before we move on, I need to let you know about my friend, Mark Pfeffer and his show,

[00:25:50] People Tech. If you're looking for the latest on product development, marketing, funding,

[00:25:56] big deals happening in talent acquisition, HR, HCM, that's the show you need to listen to.

[00:26:04] Go to the Work Defined Network, search up People Tech, Mark Pfeffer, you can find them anywhere.

[00:26:11] We've got to go back and rethink how to draw that up a little bit better. And you're right,

[00:26:17] it's like two thirds of the big AI companies are headquartered in California. So he understands the

[00:26:25] importance of whatever gets passed there. And the last thing he needed was that to become like a New

[00:26:32] York City bias law that was unenforceable and largely ignored.

[00:26:39] Your point that you said there about people being able to kind of circumvent the rules that do get

[00:26:45] and get put in place is my reasoning behind. I really haven't seen much legislation passed anywhere

[00:26:54] that I think is going to be effective or have really have much teeth to it over any like long term.

[00:27:02] And the reason is because these models are changing at such a pace. And then also the techniques that

[00:27:08] we're able to use now, such as like the ensemble type technique or mixture of experts or the router LLM.

[00:27:16] So that's like very technical terms. But the idea of that is that now someone could take and run

[00:27:23] several different models on open sourced local on their own computers and devices at their home.

[00:27:31] You're not going to be able to regulate that in the same way that,

[00:27:34] you know, it's still really difficult now. And, you know, Napster is gone. LimeWire is gone,

[00:27:40] but Torrent still exists. They were never able to squash that out because it's out in like the the open source community.

[00:27:49] And whatever gets passed here in the United States is not necessarily going to be honored or

[00:27:56] mirrored back in other countries around the world as they're trying to really,

[00:28:00] you know, make a name for themselves in this space as well. So that's why I think that,

[00:28:06] you know, the next several years we're really going to be threading the needle and almost put this more on

[00:28:12] personal responsibility and education, because I think this is a really, really difficult problem

[00:28:17] for legislators to figure out anywhere, no matter how educated they are. It's just going to be

[00:28:23] it changes too fast. Even by the time they get something passed today,

[00:28:27] GPT five or whatever the latest greatest thing could come out next week. That's going to,

[00:28:31] you know, blow through some of these, these things that they might try and put into place.

[00:28:35] Yeah. I don't, I don't think it's all bad. I don't think it's all doom and gloom. It's just,

[00:28:39] it's that's the kind of stuff that keeps me up at night when I'm thinking about that, because I'm

[00:28:44] thinking about what's the future going to look like for my kids, basically. And I don't think I'm alone in that.

[00:28:49] I think most parents would be in that same boat.

[00:28:50] Absolutely. Yeah. And I think you and I are, have different circumstances just because of the

[00:28:57] different ages of our kids. And my, my daughter's in high school, she's junior in high school already

[00:29:02] taking the PSATs later this month and starting to look at colleges and stuff like that. So

[00:29:09] that horse has definitely left the barn, but for your kids, I mean, who knows what the labor market is

[00:29:16] going to look like and the, even what they're going to learn in school when they get to, you know,

[00:29:23] you know, middle school and, and high school. I hope that the curriculum starts to adjust to

[00:29:31] accommodate that. I hope AI is incorporated into that, just like digital technology has been

[00:29:38] incorporated into the education system now. And, you know, they're using computers, they're digital

[00:29:44] natives, they're using, you know, smart boards instead of chalkboards. And I just think they can

[00:29:51] handle it, honestly. And it's a matter of how quickly we can apply some agility to the education

[00:29:58] system and get, to get teachers comfortable so that they can in turn teach the kids good habits,

[00:30:05] you know, using AI where you should not where you can, right. As a mentor, as a tutor, you know,

[00:30:11] now everyone can basically afford a tutor. So we're going to deny that to kids because you don't

[00:30:20] trust them to not use it as a calculator, to just give them the answer. Just like, you know,

[00:30:26] AI being used by candidates and in the job search process, there are scenarios where it's appropriate

[00:30:33] to use it. And there are scenarios where it's not, and, you know, there's got to be some level of trust.

[00:30:38] Otherwise, it's going to be yet another cat and mouse game trying to catch people, you know,

[00:30:44] using it inappropriately. That's one of the most hopeful messages that I've seen in this space,

[00:30:49] like period. It was about a year ago, there was a TED talk from the founder of Khan Academy. And it's

[00:30:56] only like 15 minutes, and he's basically talking about what's the future of education look like.

[00:31:00] And I found that really inspiring. And also, I practice a lot of what he talked about in that

[00:31:08] on a daily basis with my boys. So as young as they are, they already know how to interact with,

[00:31:14] if I say all the assistant's name, they might chime in here. But if I say, hello, G name person,

[00:31:20] or hey, S name lady, they know how to talk to her. And also when they're out learning, because

[00:31:25] their kids and their curiosity is just, it's, it's alive and well. So they'll see a new leaf or a bug,

[00:31:32] and they'll go get that thing or come run and get mom and dad and say, come get your phone, come get

[00:31:36] your phone, take a picture and ask, you know, one of the they call it ask the robots to tell me more

[00:31:42] about it, because they want to learn. And we'll either fire up perplexity or some of the other

[00:31:48] things to be able to take a picture. And I'll tell the system, I act, we actually use the voice on there

[00:31:53] now a lot. We'll tell the system, hey, you're talking to my son Jackson, he's about to turn five.

[00:31:59] So remember, when you're giving your response back, I'll kind of prime it with a vocal prompt like that,

[00:32:04] to explain to him what this moth is that he just caught, that like he's a five year old,

[00:32:09] but he really, really wants to learn, you know, the details about this thing. So that's, that's

[00:32:14] beautiful, because I don't know the answers to all that stuff. I can go Google it, but I can't do

[00:32:18] that very fast, not fast enough to, to quench his appetite to learn about that moth, but perplexed

[00:32:25] as you can. And one of the things that we've done in our Maven cohort is something that Amy,

[00:32:32] my wife and I have been dreaming about here for the last year or so is, could we potentially set up

[00:32:40] personalized tutors for our boys with AI agents? And we've even let them interact with Hume AI,

[00:32:48] that's a, that's a voice agent. We've done perplexity, they talk to chat GPT. I mean,

[00:32:54] I build voice agents for a living on different platforms, so they get to come in and talk to

[00:32:57] dad's robot sometimes. But connected to that, what if we could have an empathetic tutor that in its

[00:33:05] knowledge base or RAG retrieval knowledge base, it has the curriculum that we want to teach from.

[00:33:11] And it's adaptive and can teach based off of Socratic principles. So it's not just feeding them

[00:33:17] the answers, it's also trying to help them learn along the way. And it can also account for

[00:33:23] their personal learning styles, whether they're, you know, visual learners, they learn by doing,

[00:33:28] they learn by reading in further, or they learn just by going back and forth and talking to a system

[00:33:34] about it. I mean, now you have the, like you said, it is the best possible tutor. And we've built an

[00:33:41] MVP for that in our Maven cohort called the professor. And the professor is trained on all the coursework

[00:33:47] that we're going through. It's trained on a lot of, you know, stuff that we throw in its knowledge base

[00:33:52] about AI, about the tools that we're using, and even all the transcripts from our meetings and lessons

[00:33:59] get chucked in there as well. So that the students in the course can actually go and interact with the

[00:34:04] professor to get their answers. And it is adaptable to their learning and speaking style, like how they

[00:34:10] choose to learn, how they'd like to be talked to. So I see that it's, that that's been going really

[00:34:15] well in there. Now, not to say that it hasn't, you know, misfired here and there, because this is tech,

[00:34:20] and that's an MVP. But yeah, I think that figuring out that problem is really hopeful for kids of all age,

[00:34:28] because, you know, even your child, that's going to be having to think about career here soon of like,

[00:34:34] what do they want to do when they get big? Or when they when they get older is that's a that's a tough

[00:34:40] question to answer now. And it's probably never been easier to be self employed, if you have that

[00:34:48] drive at all than it is now. Because you're able to do things as a solopreneur or a very small agile

[00:34:55] team that just was not possible before. Yeah, no, absolutely. I can certainly appreciate that myself.

[00:35:02] So you answered some of my questions I had for you as as we went and as we were talking, one of them

[00:35:09] being like some of the projects that you're working on sounds like you had some personal

[00:35:13] projects that were pretty cool for your kids and for the family. And then of course, we talked about

[00:35:20] responsible AI in depth. But you know, maybe some of the projects you've done for clients would be

[00:35:27] interesting to hear about, especially, you know, if it's around like the, like your Maven course talking

[00:35:33] about like, you know, workflows and how to connect some of these, you know, sort of disjointed pieces so

[00:35:39] that you're not just doing, you know, task by task, but you're, you're sort of stringing them

[00:35:43] together. Can we like unpack one of those? Sure. Yeah. I have to talk about them a little bit,

[00:35:50] not all the way in the details, just because of like NDAs. But when I always love talking about

[00:35:54] this as best as we can, because I think that's what helps people have those light bulb moments of

[00:36:00] what you can do now. And recently, like you said, at the top of the podcast, agentic workflows are the,

[00:36:06] the flavor of the day right now, which is basically us building up AI agents or assistants.

[00:36:14] Because, you know, not all assistants are agents, but not all agents are assistants. Like it's part

[00:36:19] of the jargon that gets thrown around through there that gets kind of, you know, mixed, mixed up. But

[00:36:24] we are working with an organization with a bunch of, they're a fractional C-suite team. And I'll just

[00:36:33] kind of leave it top level like that. They go into organizations, they're offering C-suite level

[00:36:38] services. And everyone on their team is later in their career, they're full of wisdom. They're really

[00:36:46] highly regarded folks in their, their industry. And even on their own team, they had some real doubt

[00:36:54] of if it was possible for us to build an AI system that could emulate their processes to the same level

[00:37:01] of quality and standards that they could do themselves. They just, they, there were some

[00:37:06] believers and non-believers on the, on the team there. And essentially what the, what the workflow

[00:37:12] is that they wanted to set up is when they hire, or when they bring on a new client, they'll go through

[00:37:18] this discovery phase where their clients will send through all of these different files into a drop zone

[00:37:24] folder. And their team then has to go fine tooth comb through every single one of these files. They

[00:37:30] could be PDFs, documents, PowerPoints, financial statements, organizational charts, like literally

[00:37:37] anything that you can think of. They're going to chuck that into this drop zone folder. And this

[00:37:42] fractional team goes through all that. They're making their notes of what's important or what's relevant

[00:37:48] to their engagement. And that's extremely time consuming for them. And it's also, you have to

[00:37:55] kind of have this head knowledge that they've built up over their entire career of what could potentially

[00:38:01] be relevant in this engagement based on this just random array of different files that we're looking

[00:38:08] at. They wanted to see if we could set up an agentic workflow that could help them with that process.

[00:38:14] And we've been able to do that where those files can now still hit the drop zone folder that goes

[00:38:21] through a team of agents that will review each of those files. We've got it set up right now for just

[00:38:27] five. That's enough. So five different agents will look at the same file. They all are writing their own

[00:38:34] summary or extraction of what they think is relevant or could be relevant for this fractional team.

[00:38:39] And then they pass along that summary to a different agent that's going to review all of that work.

[00:38:46] And then it will then make a summary of the summaries. Beyond that, there will be an evaluator step.

[00:38:54] So this is kind of like, if you've heard of the chain of density framework for chain of density

[00:39:00] summarization, where it's taking and writing a denser and denser summary, that's definitely helped

[00:39:06] influence the way that this works. And at the end result, our goal is to have a one page summary

[00:39:13] that is written that anyone on this team can pick up without having any other knowledge about this

[00:39:19] organization or the project at hand and be able to know all the details that are most pertinent for

[00:39:24] them and the type of work that they do related to this potential client. And I will say that we've

[00:39:32] done that and per their own evaluations from this team, they gave it marks of 11 out of 10. I think

[00:39:38] the worst evaluation that we got from them was a nine out of 10. And I'll take that. That's still a B.

[00:39:44] So, but they really didn't think it was possible. And it even went as far as it was able to

[00:39:51] take things from these files that weren't even explicitly said on the page and do reasoning and

[00:39:58] thought processes behind it that the human team would have done. And one example is there was an

[00:40:05] organizational chart that ran through this. And on this org chart, the number two person on the org

[00:40:12] chart said acting on it. So it was just like, it looked like a regular organizational chart had

[00:40:16] everyone's name, their position and the department that they worked in. And that's it, nothing else.

[00:40:21] But this, the number two in charge said acting. And in the summary, they got sent to the fractional team.

[00:40:28] It made a comment, and I'm really going to not word this as well as eloquently as it did. But it said

[00:40:34] that since this person is now in the acting position, that shows that this organization that you're working

[00:40:41] with is in a transitional period. That means or could mean that they're going to be much more open to

[00:40:47] additional changes in the work that you all are going to be performing for them in their engagement.

[00:40:52] So nowhere on that page did it do that. But a seasoned person on this fractional team would have

[00:40:59] seen and recognized that same thing. In fact, the CEO of that organization was like, my lead person

[00:41:05] who would have done that is, I don't even know if they would have caught that necessarily and written

[00:41:10] that into the summary. And all of these, these one page summaries that are made during this process

[00:41:16] are then stored in a rag retrieval database, where a final summary will happen right before they go back

[00:41:25] and do their proposal of like, here's, here's the fix. Here's what we think that we're going to be

[00:41:30] doing for you in this engagement. Because that's that emulates what their their actual team does. They

[00:41:36] they go through this process, they have all their meetings, they do their discovery, they're basically

[00:41:40] doing their their analysis of where y'all stand currently, as we're doing this consultation for

[00:41:45] you. And where do we think that you can get to? And how do we think that we're going to get you there?

[00:41:50] So normally, their whole team would just be doing this. But now this agentic workflow does the process

[00:41:56] of going through all the files, they can still go in and read them themselves if they need to like

[00:42:00] double click into it. But it went from this process that takes weeks into, I mean, literal minutes,

[00:42:09] if you have all the files right out the gate, because it just runs through all of them, it creates that

[00:42:14] summary. And from there, it creates this plan for what is the fractional teams, potential services that

[00:42:22] they could offer. And it even goes into depth of like, how would we they go about doing that? So it's

[00:42:28] helping them formulate their plan for their their pitch or project plan for the their their client

[00:42:33] that, you know, I know I'm having to speak vaguely about that. But that one was, I was particularly

[00:42:40] impressed with the reasoning that our team was able to program into that. Because these folks are top of

[00:42:49] their field. And I know that we, we can get these models and these these agents to be able to reason

[00:42:55] really well, but to be able to emulate someone that is, you know, 60 plus years old, that is just

[00:43:01] really, really full of not just like industry knowledge, but just life wisdom, and to be able to

[00:43:08] deduce information out of just a simple summarization flow at the same level that they could, if not

[00:43:14] better, that surprised me. And I think that goes to show that knowledge work is about the cost of

[00:43:22] whatever the average token rate is now from, from Anthropic or Chat GPT, whatever it costs per

[00:43:30] million tokens is probably what your head knowledge is worth these days.

[00:43:33] Yeah, no, I definitely tell people that this is not just about, you know, road tasks and drudgery.

[00:43:41] This is, you know, knowledge work is affected as well. I'm not saying we remove knowledge workers,

[00:43:49] as well as these things advance, but you've got to really understand how to best position yourself and

[00:43:54] your own human durable skills, right? Because as AI takes on some of some, some of these other

[00:44:01] tasks that were previously the domain of, of humans and human cognition, human creativity,

[00:44:08] you know, taking on some of those things. But so the more we can think about how to,

[00:44:12] you know, reposition ourselves and the value that we create, the better off we are. So it sounds like,

[00:44:19] I mean, in that particular case, you're saving a tremendous amount of hours because it's not,

[00:44:25] you know, those, those resources, you know, fractional or not are experienced and tenured,

[00:44:31] and they've got the domain knowledge, they've got the, the work experience, they've got a lot of value,

[00:44:37] but they're also expensive resources, right? So I don't know if you can share anything about the,

[00:44:44] you know, the outcomes, but I imagine it's saving them, you know, significant, both cost and time.

[00:44:50] Hi, I'm Steven Rothberg.

[00:44:52] And I'm Jeanette Leeds.

[00:44:53] And together we're the co-hosts of the High Volume Hiring Podcast.

[00:44:57] Are you involved in hiring dozens or even hundreds of employees a year? If so,

[00:45:01] you know that the typical sourcing tools, tactics, and strategies, they just don't scale.

[00:45:06] Yeah. Our bi-weekly podcast features news, tips, case studies, and interviews with the world's

[00:45:12] leading experts about the good, the bad, and the ugly when it comes to high volume hiring.

[00:45:18] Make sure to subscribe today.

[00:45:20] It definitely is. And it's helping them. Their reasoning behind this is that they're a small

[00:45:26] team and they wanted to be able to scale because they're, they're very much so on the side of the

[00:45:32] demand is much greater than what they can currently supply. And they were hoping that layering more

[00:45:39] AI into their work would be able to help them scale that. And it, it definitely has. They've,

[00:45:45] they've taken on much, well, they already had several large clients, but now they've taken on even,

[00:45:50] even more larger clients. We even had a meeting with them yesterday just to go over some different

[00:45:55] things. And they were sharing that they had just brought on two additional clients.

[00:45:58] And they're, they're feeling like they can do that because, you know, now they have these,

[00:46:04] this AI team in place with them as well to help them with these processes that, you know, before

[00:46:10] they, they wouldn't have been able to do it because the man hours is just not possible. And I also want

[00:46:15] to like put in the caveat that when we design the system for them, in my opinion, every really good

[00:46:22] business has some sort of a secret sauce. And so part of our job, when we're designing this and writing

[00:46:28] the system instructions and the, the workflow and all that sort of stuff for these different processes

[00:46:34] is that we'll go through some really intensive discovery phases ourself, but we're trying to

[00:46:40] understand who they are as people, what they're inspired by, what their personalities are like,

[00:46:46] how they, how they work through and think through problems related to their, their work. And also

[00:46:51] just personally as well, what's the type of information they have in their, their company

[00:46:56] knowledge base, like not AI, just like in general, what are the types of things they're trying to learn

[00:47:00] about? And what does this process, we'll go through this process mapping exercise because a lot of

[00:47:08] companies do not have SOPs, especially in small and medium-sized businesses. Some of them don't even

[00:47:12] know what that acronym even means. And so we just do this really fun zoom call or in-person meeting and

[00:47:19] we'll record it. And if we can just talk to them conversationally, like most business owners or

[00:47:24] leaders in a team can walk you through a process verbally of like, oh yeah, we do this. And then

[00:47:29] Bob over in accounting, he does this part. And then Tyler over in marketing, he does this bit.

[00:47:33] And they're literally verbally writing the SOP for us. And we have a workflow on the backend that will

[00:47:39] take that transcript, turn it into SOPs. We also get a lot of information out of them and those

[00:47:46] transcripts to be able to use that to influence these very personalized assistants that are in

[00:47:53] this workflow that emulate how they do the job. So it's not just the bone stock, Claude or chat

[00:47:59] GBT that's doing that, or even just some more generalized assistant at that task. It's, it is that, but also

[00:48:05] personalized very much so to their business, the way that they do things and their thought processes

[00:48:11] behind it. And that's how I think we've been able to get the really high quality results out the other

[00:48:15] end that, that passed the sniff check of did a person do that or not? Wait, so, so when you created

[00:48:21] those five agents who were going to look at those documents, was each one customized to five

[00:48:29] different, you know, experts, but came in and came at it with the different through their own digital

[00:48:35] lens. In other words, were they, were they five sort of digital twins of, of the people that would

[00:48:41] otherwise be going in there themselves to review that? That's what I wanted to do. I personally wanted

[00:48:47] to do that to get very diverse backgrounds and points of view, because that's getting into like

[00:48:54] the, the mixture of experts and like chain of thought type reasoning that we could do with that.

[00:49:00] But our, our client wanted to emulate their CEO and leader because he is the real serious brainpower

[00:49:10] over there. And he's the one that's taught a lot of his team members how to do these different

[00:49:14] things. So they wanted to make these different assistants really emulate him and his processes.

[00:49:20] So the assistants are, are modeled after him and the way that he does things on top of just how they

[00:49:28] do things as an organization to all of that's kind of baked in, but you, you nailed it. Like I really

[00:49:33] wanted to look at like, here's these five people on their team. We're going to make a, a digital

[00:49:39] assistant that knows exactly how they like to work and how they view things so that you're really getting

[00:49:44] the same type of result that, that they might be able to produce.

[00:49:47] This could go one of two ways, right? One is listen, that this is one of the enormous benefits of,

[00:49:55] of creating custom GPTs or, or letting one of these generative AI tools really understand

[00:50:03] you and how you think and how you write or whatever. Cause otherwise you're going to continue to get

[00:50:07] generic sounding responses. You could ask different, you could ask chat GPT, you could ask Claude,

[00:50:12] you could ask, you know, Gemini or whatever. And they all, a lot of the output starts to sound the

[00:50:17] same. Well, it won't, it'll sound more like you. It'll sound more personalized. It'll take on your

[00:50:24] writing style, your tone or whatever. If you customize these solutions, um, which you get vastly incremental

[00:50:31] value from, but we also have to acknowledge that the very fact that it can do that and basically create

[00:50:39] a, a sort of, at least some subset of a, of a digital twin kind of concept that in some ways people are

[00:50:47] going to feel like some of this technology is, is basically they're, they're training their AI

[00:50:53] replacement. And so I think we just need to acknowledge that, that sort of delicate, you know, balance and,

[00:51:00] and also why being responsible by design. Uh, and as we use these tools is just crucial for positive

[00:51:09] developments in this area.

[00:51:11] A hundred percent. Like their, their goals in this were that they wanted to be able to scale beyond

[00:51:17] their current capacity. And just like any business, it's really difficult to find team members that not

[00:51:25] only match your culture, but the educational background and the domain expertise, all that sort

[00:51:31] of stuff. And then you have to, once you find those really amazing people to bring into your team, keeping

[00:51:37] them on your team can be even more of a challenge because those type of superstars are generally are

[00:51:43] going to get poached from other companies, or they're going to go off and start their own businesses

[00:51:46] or something like that eventually. So knowing that, uh, where they are in their career as well, because a

[00:51:53] lot of these folks on this particular team are later in their, uh, career, they knew that they wanted a

[00:51:59] better work-life balance, but they also have a certain lifestyle that they wanted to maintain.

[00:52:04] And so to be able to maintain that, you know, that would require a lot of man hours. Now they're able

[00:52:09] to either stay on par or pace with their growth that they've been experiencing now without having to

[00:52:15] put in all those additional man hours and still keep their workload manageable and the quality, uh,

[00:52:21] to the level that they would expect to do if they were doing it themselves. The other really interesting

[00:52:26] thing that was fun about this project is it wasn't in the United States. This was with a company in New

[00:52:31] Zealand. So thank God they spoke English, but they speak very different English than we do. And one, one of

[00:52:38] the fun things that, that I don't have the life experience to be able to gauge when I was doing my evals on

[00:52:44] these was this good or not because we had it in, in the responses and in the way that it would talk,

[00:52:51] it would sound like them. We did a brand voice analysis on it so that it would sound more like

[00:52:59] them or, or the tone and style that they wanted it to be. And it spoke New Zealand English or the

[00:53:05] Queens English or whatever. And it even uses like cloak colloquialisms and phrases and vernacular that were

[00:53:13] only in New Zealand. Do you speak this way? And they probably think the same way when they were

[00:53:17] talking to me during the project, I use a lot of phrases from the South. They're like, what the

[00:53:21] heck are you saying? And I literally would have to, we would stop and ask each other throughout this

[00:53:25] project. Whoa, whoa, whoa. I don't know what you just meant. Can you, you're speaking English,

[00:53:29] but I don't understand it. Well, we had their assistants speaking New Zealand English, and I'm the one that

[00:53:36] wrote that from the United States, these LLMs are trained a lot on information and data from sources in

[00:53:44] the United States, because a lot of the companies are based here. And the fact that it was able to

[00:53:49] translate that and mimic their speaking style and use metaphors and verbiage that is common for them, I

[00:53:58] thought was also a really cool and fun win there that it did a really good job at that part of the task

[00:54:05] too. So that's, that's where we're getting more towards this idea of a universal translator too, which is

[00:54:13] also, I think very helpful because now it really does open up this international market even more than

[00:54:18] we already have had with, with the internet. No, absolutely. You know, one of the things I wanted to, you know,

[00:54:24] throw out there for the listeners, like when we talk about, you know, agentic workflows, and you mentioned

[00:54:31] this gives like solopreneurs like myself and, you know, small businesses, some, some superpowers in

[00:54:39] the sense that I don't necessarily have to take it all on myself, right? Like it's, there's only so many

[00:54:45] hours in the day and you need to maintain your, your sanity and your work life balance a little bit

[00:54:52] of a degree, but you also don't necessarily have to go and find, you know, a contract resources or even,

[00:55:00] you know, a virtual assistant or an intern. I mean, I, I still contend that could probably use one, but,

[00:55:05] but still there's, there's AI out there that can help me if I take the time to sort of put the pieces

[00:55:15] together. So, I mean, it sounds like you either through, you know, light magic or galactic ranch,

[00:55:21] you know, could help someone like me with some of the, what I think would be good agentic workflow

[00:55:27] use cases. You and I have talked once before about like the production and, and publication of,

[00:55:33] of this podcast, right? I mean, I, I do, I do research. I understand the guests. I think about

[00:55:40] how to frame the discussion. And then I take, you know, I've got show notes. I've got the transcript,

[00:55:47] I've got a summary of the transcript. I've got all those keywords. I need to create an introduction

[00:55:52] to the podcast based on the summary. I've got to create a social posts to promote the episode.

[00:56:00] And I'm in and out of like 10 different tools. So it's not just the number of tools and perhaps

[00:56:07] some of those tools may not be free. It's not just that, but the, you know, each time it takes,

[00:56:15] you know, me as the, as the human in this workflow and the loop to go in and, you know, copy paste

[00:56:22] or whatever. I'm, you know, in notion, I've got an AI in Google docs, I've got Google Gemini,

[00:56:27] I've got the chat GPT plus perplexity that I use for research. I mean, I'm all over the place, right? So

[00:56:36] there's gotta be a way if I, again, if I put in the time to structure it and work with, with someone

[00:56:42] like you to put those pieces together, use something like, like make or whatever, which is sort of like a,

[00:56:49] like a Twilio or a, you know, one of those connector API connector kind of orchestrator things.

[00:56:56] But how do you like literally put all the pieces together? So if you put in that work up front,

[00:57:00] it could save you a lot of time in going task switching and application switching and, and things

[00:57:09] like that. So if that takes me, I don't know, let's just say it takes me, you know, eight hours to do

[00:57:15] all of that. How much time could I save? And that gives me time to talk to other prospects and do other

[00:57:21] work so that I'm not working 12, 15 hours a day or whatever.

[00:57:24] Yeah, for sure. It's not something that happens overnight, for sure. That is something that it

[00:57:30] takes time to identify all these different tasks and processes that we're doing as people in our work

[00:57:37] and then start to figure out where does it make most sense? Where am I going to get the best bang for my

[00:57:42] buck on starting to, you know, hand that off to these AI systems a bit more to be able to help. A couple of the

[00:57:49] the biggest things that I've seen to help me just as like really easy wins are if you can begin to

[00:57:56] create whatever you want to call it, a second brain system for yourself or your business or both,

[00:58:02] where you can consolidate all this data that you're creating. People don't even necessarily think of it

[00:58:07] as data. It could be, you know, we have transcripts from all these different zoom calls and team

[00:58:13] meetings and all these sorts of things that we're in on now. I have all this internal documentation

[00:58:18] that we create when we're working. We have that land back into the second brain system. Our CRM,

[00:58:24] depending on if you have that separate or in the same location, that's something that you

[00:58:29] would want to have in that second brain system. I personally use Notion for that. That's just a

[00:58:35] personal choice, but a lot of our clients, they don't use that. They'll be in the Microsoft or

[00:58:40] they'll be using things like HubSpot or Salesforce on top of a bunch of other usually disconnected and

[00:58:47] siloed tools that have all this really powerful data in it. And when you're able to start connecting

[00:58:53] and consolidating all that, or at least making the connectors so that these agents and your AI

[00:59:00] workflows or AI team members, whatever it is that you want to call them, can access some of that stuff.

[00:59:07] That's when you're really able to start connecting the dots there and see some really powerful results.

[00:59:12] Like as an example, to even talk about what you were going through there just for editing the podcast

[00:59:18] and all the different processes in that, us launching this cohort on Maven, Sara and I have been working,

[00:59:26] I've spent a lot of time in the studio here recently recording lessons, making the lesson plans,

[00:59:33] making the copy that's going to land on the Maven page, making the knowledge-based items that need

[00:59:40] to go into the professor's brain in the rag retrieval knowledge base, creating the projects and the

[00:59:46] homework, actually filming and editing this stuff. Like that's a lot of stuff. The other day I was joking

[00:59:52] that I had, I have a dual monitor set up and I had split apps in both of the monitors. So I had four

[01:00:00] screens up basically that I was looking at. And in two of them were Cassidy automations running in the

[01:00:07] background on client projects that I needed to get, you know, stuff moved forward on while at the same

[01:00:13] time on this other screen was completely dedicated to work that I was doing for this cohort. And I had

[01:00:19] Descript open and then I had Typing Mind open where I was in there just going back and forth, having it

[01:00:24] create these different assets like the lesson plan and all that kind of stuff. While Underlord was over

[01:00:29] here in Descript, you know, shortening word gaps and, and figuring out what are the hottest

[01:00:35] clips or takes that I could use to repurpose for, for marketing this thing. I would never be able to do

[01:00:41] all that stuff if I had not taken the time and, and really outsource the information from my head and

[01:00:51] into a system so that I can then begin to teach and give access to this knowledge that lives in my head

[01:00:58] or these, these, all these other independent places so that these AI assistants could do it mainly in Cassidy and

[01:01:05] typing mind is where those assistants were doing it. The Descript agent was just focusing, you know,

[01:01:09] obviously on the video that we were editing, but it's, it was a really cool moment. And also I had to

[01:01:16] like kind of zoom out on myself and be like, this is like a mad scientist moment here for me in my office,

[01:01:22] just trying to conduct the symphony of AI agents on the back end to get all this work across the finish

[01:01:29] line. And I was able to do it. That's how I'm able to be able to pursue so many of these different

[01:01:35] projects and passions that I'm in is because I have slowly, but surely began to look at what is it that

[01:01:42] I do on a daily basis and what bits and pieces of that make the most sense to be able to, to start

[01:01:48] offloading to my AI team versus me, or we do it collaboratively one or the other. You know, this is about

[01:01:54] what else could we be doing when we know some percentage of our day is of everyone's days doing

[01:02:01] sort of administrative things. I mean, even if you have, even if you're fortunate enough to have an

[01:02:07] actual assistant, there's always other things that could be more routine and could be automated. And then

[01:02:13] how do you use, move up this sort of value chain of some of the AI capabilities to augment what you do.

[01:02:22] And some of that is, is just summarizing things that you need where the volume of, of text and

[01:02:29] information is just overwhelming and you need a trusted sort of partner to summarize that. I like your

[01:02:35] approach that takes multiple agents and pulls that together because a single summary may overlook,

[01:02:42] may miss some context and, and overlook some things. So I think that's a really interesting

[01:02:47] approach. It's almost like applying some of the concepts of collective intelligence and cognitive

[01:02:53] diversity. But within, you know, the, the AI domain, I also just want to point out, like, I'm not trying

[01:03:02] to put my business on autopilot, right? Nothing. I don't want anything to be posted on social media or

[01:03:11] anywhere else without me having my eyes on it. I think AI has been great to draft things, but we're still

[01:03:20] at the point where, and part of this is about the customization and it learning my writing style and

[01:03:26] tone or what have you. But, you know, so far I've, I have edited everything that it has created for me,

[01:03:34] whether that's emails or social posts or, you know, the summaries of, of the podcasts or whatever. There's

[01:03:42] always a little bit of tweaking that needs to be done. And then just almost, you know, proofreading,

[01:03:47] you know, what, what it's done. So, but it still takes a lot of time away from things that I would

[01:03:55] otherwise have to do and, and time is your most precious resource. So what are you going to do

[01:04:02] with the time savings and how you're going to reinvest that in yourself and your organization

[01:04:08] and learning more about some of these topics that are influencing the future of work?

[01:04:13] A hundred percent. Very well said.

[01:04:15] So Tyler, I know we kind of covered a lot of ground over this conversation. Just in,

[01:04:21] in closing, I mean, any, any final thoughts and advice for, for listeners in terms of,

[01:04:26] as the title says, uh, elevate your AIQ?

[01:04:30] I mean, no matter where you are in your, your headspace of where you are in your AI journey,

[01:04:36] if you want to call it that there's no time like right now to begin just practicing and testing with

[01:04:42] this stuff and, and learning and finding trusted resources that you can really figure out what are

[01:04:49] the frameworks and the techniques that you need to do to get these AI systems to work even better

[01:04:54] than just going in and just talking in plain language to chat GPT. A couple of the really key

[01:05:00] things that I think that anyone can start with are as simple as using something like a, an otter or a

[01:05:07] fireflies or a fathom. You want to make sure that, uh, whatever it is, the security layer is in place,

[01:05:13] like SOC 2, type 2. I know that's something we talk about a lot. All of those are AI note-taking apps.

[01:05:18] And being able to just that one simple thing of having an AI show up into your meetings so that

[01:05:24] you can be 100% present in the conversation and have everything captured. So you have perfect recall

[01:05:30] from that. And then the next step is that if you're able to have all that stuff re-sent into a

[01:05:37] repository, like a second brain system, just that one thing is incredibly powerful. So now as you

[01:05:44] begin to learn how to connect in, you know, chat GPT or even, even further down the road, you get into

[01:05:50] agents or assistants now can go back and look through all these different meetings that you have

[01:05:55] to be able to answer questions of like, Hey, Bob and I met, you know, six months ago. Uh, and we talked

[01:06:00] about this, that, and the other, can you refresh my memory on all that? Um, so that I'm up to speed or

[01:06:06] if this was a project, what was the thing that, uh, Bob and I went and talked to client ABC over here

[01:06:12] about, I was supposed to be doing something and I can't remember. Can you give me my complete to-do

[01:06:15] list? That is super helpful for everyone. And it's a really easy thing to set up and, and get into play.

[01:06:24] It doesn't any, anyone can pretty much do that. The other thing that I like using a lot that is also

[01:06:30] just a really easy lift is using a tool such as audio pen, or there's voice notes, or even the

[01:06:37] fireflies app can do this where you can just record a voice memo. So I have been embracing the second

[01:06:44] brain mentality for a while now before AI, where I'm trying to offload things out of my cognitive load.

[01:06:51] So I free up and can stay in flow state. And I need to remember this stuff and get it down,

[01:06:56] or I just won't remember it anymore. I've got dad brain now, literally. And so if I don't get it out

[01:07:01] of my head and into a system, then it'll be gone. And being able to use one of those type tools to

[01:07:07] just very quickly, just open it up, vocally brain dump into it, have that transcribed. You can also

[01:07:14] begin to set up automations to turn it into whatever downstream, like an email or blog posts.

[01:07:21] The world's your oyster there, but have that again, land into that second brain system. So now we have

[01:07:27] my thought process, whether it was just like a quick thing I needed to get down a to do list item

[01:07:32] coupled with also here's all my meetings that are now being recorded and added into there.

[01:07:36] That's a lot of your intelligence and what you're doing in a day to day basis that pre AI trying to

[01:07:44] keep up with all that and remember all that kind of stuff is I'm not good enough to do it.

[01:07:48] As I've tried all the different productivity systems like Eisenhower square, and then even like the

[01:07:55] Tiago Forte's system or the PARA method, all those types of things, and I struggle with it. And now that

[01:08:03] I'm able to still keep the concept of centralizing that information, but then connecting AI to it, I can

[01:08:10] outsource that out of my head and into the system. And now I can literally have so much stuff in there

[01:08:17] now, both personally and professionally. It probably knows me better than I do. I can ask it questions

[01:08:22] about myself and it'd be able to tell me things back that I might not even recognize about myself

[01:08:26] or that's going on, which is, it's pretty cool when you can do that.

[01:08:30] Awesome and scary.

[01:08:32] Yes, awesome. Both can be true.

[01:08:36] Yeah, yeah. No, you've got me thinking about, I have this giant pile of post-it notes with ideas and

[01:08:43] follow ups and to do's and whatever. And it's like, do I just use one of those transcribing

[01:08:50] tools and just start reading them into it? Or do I take a picture of all of them and

[01:08:56] you know, upload it and will it recognize my handwriting? Like I gotta figure out something,

[01:09:02] but I'm tired of walking around with two inch thick pile of post-its over the last couple.

[01:09:06] Yeah, it depends on your chicken scratch. Like mine, mine's pretty rough. So that's why I just

[01:09:10] brain dump into it, but they can do that. You can just take pictures of it. I even learned a hack from

[01:09:16] one of my buddies, Matt Burton. He went and took a picture of his bookcase and it was able to extract

[01:09:22] the titles off of every single book in his personal library. And then he had all those listed out as

[01:09:29] books that he has read and added that sort of information into a second brain system. So now these

[01:09:34] agents kind of know like, okay, what else does Tyler know about? What is he interested in? What

[01:09:39] has he read? Because all these things, whether we recognize it or it's still just going on in our

[01:09:45] unconscious minds that affects us and our thought processes and our level of intelligence. So getting

[01:09:53] our AI systems up to speed on that kind of stuff is, that's the type of information that I'm throwing

[01:09:59] into my second brain system. That and recipes and social media posts and all that kind of stuff,

[01:10:05] I'm getting it all in there. You can have a fear for AI, but make it a healthy fear. Don't be so scared

[01:10:11] of it that you're worried about Skynet, that you're going to miss the boat on how much of a positive

[01:10:15] effect this can have on your life, your business's life and your family's life. Because it definitely

[01:10:21] does. And it definitely can. It's not all bad. In fact, I would argue that the net effect of all this

[01:10:27] is going to be that we're going to be able to solve in the very near future. A lot of these very,

[01:10:32] you know, unsolvable problems such as like curing cancer. How do we feed the world so that no one

[01:10:38] goes hungry anymore? How do we make sure that everyone is taken care of with employment and

[01:10:44] housing? Like again, those sound like, you know, this sort of picturesque future that is not possible.

[01:10:51] But I really truly believe that this could be the technology that could help us achieve some of

[01:10:59] those things. I agree. I think to your point, these previously sort of insurmountable kind of challenges,

[01:11:04] these super complex, you know, global challenges, you know, between our collective intelligence and

[01:11:11] artificial intelligence, it seems like there's a lot of opportunity to address, you know, some of

[01:11:17] these, some of these global issues. So that is a whole nother conversation, the humans plus AI to solve

[01:11:25] world hunger and world peace. So Tyler, we're gonna, we're gonna have to leave it there. But this has been

[01:11:32] awesome to really get into the weeds with you on some of these projects you're working on. And

[01:11:37] you've given my listeners some amazing advice. I know I've got a bunch of things I need to now go

[01:11:43] play with and try to improve my own, my own workflows. Always good to talk to you. And thank

[01:11:50] you for spending so much time with me today. You as well. Hey, thank you so much. I really enjoyed it.

[01:11:55] Excellent. All right, everyone. That's gonna wrap it up for today. Thank you for listening to another

[01:12:00] episode of Elevate Your AIQ. And we'll see you next time.