How to Get More of Nick Berry and tBOJ

If you enjoyed this episode, please subscribe to the podcast on your favorite platform and leave a review. Your feedback helps us bring you more valuable content.

Follow us:

• Nick on LinkedIn
• Nick on X
• tBOJ Podcast on X
• Redesigned.Business

The Business Owner's Journey Podcast host: Nick Berry
Production Company: FCG

[Image Gallery: American Entrepreneur Nick Berry]

Episode Transcript for: Jake George: Why Most AI Implementation Fails and How to Get It Right

[00:00:00] Jake: they noticed that the AI spend for that, because it was the only thing he had built so far, it was, like, over $2,000. And so they're like, "Okay. We gotta pump the brakes on this.

[00:00:08] Jake: This is not worth that amount of spend." So then it's like they're asking him to fix it and add more stuff to it and whatnot, and he's just like, " I don't wanna be a developer. Like, I didn't sign up for this. This isn't even my job role. You guys, you know, nominated me for this." So- Then we come in, they had mentioned, they're like, "Yeah, we [00:00:25] tried to use AI, it's just way too expensive."

[00:00:27] Jake: And they explained it, and I was like, " should definitely not be that expensive. Something went terribly wrong."

[00:00:31] The Business Owner's Journey. I'm Nick Berry, and I've got real business owners telling their real stories, sharing their real lessons and strategies so you don't have to figure it all out on your own

[00:00:44] Nick: Jake George is the co-founder and technical lead of Agentic Brain, the AI consultancy [00:00:50] behind over seven hundred AI employees for two hundred and forty-five different businesses. Jake is anti-hype, and he said that he spends most of his time telling people when not to use AI because most of it fails because it was never built around the actual workflows.

[00:01:04] Nick: So he started Agentic Brain so they could begin with your goals and then identify your challenges and then build AI employees with a proven process

[00:01:12] Nick: Expect to learn how to tell whether your [00:01:15] business is ready for AI, when you should leave AI completely out of a process, what drives runaway AI costs, the best and the worst ways to split work between AI and your team, how to find hidden revenue in data that you already own, and Jake's take on why outsourcing your thinking is what gets you replaced.

[00:01:31] Nick: Enjoy this episode with Jake George

[00:01:34] Jake: when the models were, like, a lot older and whatever, it was just like, "Can AI do this? Is it able to do this? How [00:01:40] would it do this?" And now we're at the point with the models that it's less about like, can it do it, what model to use?

[00:01:45] Jake: Like, they

[00:01:45] Jake: could do almost anything that you could do on a computer and give it tools to do. And it's more about like, does it actually work effectively? Does it, complete the goal or the outcome that you're looking for? And then also, is it actually efficient? Or it's like, are you creating, like, security vulnerabilities, or is your spend going crazy off the rail?

[00:02:03] Jake: So I think now it's more [00:02:05] about, like, optimizing it. It used to be more of just like, just throw AI everywhere, just like AI for this, AI for that, and everyone is just kind of betting on like, oh, it's this, like, God robot sorta thing, and it's just gonna be way smarter than everyone else. It's gonna automate everything.

[00:02:18] Jake: And then that kinda fell flat on its face. A lot of companies just blew through their spend. And then it's more of just like, okay, this can work, but how do we optimize this? So, like, a lot of the stuff we see falls into maybe, like, [00:02:30] three or four categories, I'd say. that I just mentioned, which is, like, unsecured data, which I won't get to.

[00:02:35] Jake: That gets very, very technical. I'll, you know, kind of breeze over it a little bit in a way that makes sense. also just like, pissing off, like, your staff or customers, kind of the same hand in hand of just, like, throwing AI everywhere. You're not really thinking about it. What we like to call the proper implementation is intentional AI rather than, you know, as opposed, let's say, unintentional, where you [00:02:55] just slap it on everything, which is a terrible way to go about it.

[00:02:57] Jake: human in the loop, this gets skipped over a lot. And then the main one that everyone talks about, you see all the headlines, are just kind of the runaway costs. It's just like, you know, you put a card on there, you let the agent do whatever, and then you're hit with, you know, this, you know, t- $5,000, $10,000 bill or whatever at the end of the month, and it's just like, what, what even got done for that amount of money?

[00:03:18] Jake: one of the main ones that I, [00:03:20] I'd say that we see the most is that people don't often differentiate, They think it has to be one or the other, like fully autonomous or fully human do it. And there's most workflows, I would say are, of workflows are in between actually, where it's just like AI does five steps, then someone, like, checks it over and approves it.

[00:03:39] Jake: Or it's like maybe there's a step that needs to be done by a human. AI kinda does the rest. You give it a final review and send it [00:03:45] off. And that's where I'd say like, like I said, probably 90% of projects where most of the efficiency is set in. And a lot of people they're like, "Okay, well, I'll just have an agent do it all."

[00:03:54] Jake: And then you get these, you know, horror stories of it's, it's sending out the wrong things to the wrong customers, it's leaking data, it's filling out these documents wrong and just submitting them, and it's, you know, legal headache and whatnot. But really kinda the thing to do is if you break it down- [00:04:10] Take a step back and think of like, A lot of workflows, it's like they're, you know, 10 to 20 different steps if it's any sort of, you know, something you actually are using a human for that's not already automated. So you gotta look at which one of those steps or which few of those steps could you actually just have an AI do? Where does a human jump in for approval?

[00:04:28] Jake: And then kinda just, you know, push it through to the end and have someone look it over. And this is a part, you know, a lot of people [00:04:35] skip. That's where we get all this AI slop. people just kinda put out whatever the AI puts out. They're just like, "This is good. Submit it, send it along." And then on the other side, if they're saying like, "Oh no, AI couldn't do this because a human has to review it."

[00:04:47] Jake: And it's like, well, that's fine, but it could do all these first parts here.

[00:04:50] Nick: in those situations, it seems to me like chances are people are thinking that they're gonna be able to hand some things off and not have to think, and then they, they find out that [00:05:00] really the only part of it that they can't hand off is the thinking part.

[00:05:03] Nick: the copy and paste and the, the mindlessness part of it, that's what AI can t- usually take over in those situations

[00:05:10] Jake: that's exactly right. and that's why I think it gets such a, a bad name all the time, and people are starting, you know, people hate a lot of AI output. But I think it's really because it's more so the people using it, that they're just sort of like pawning off their thinking.

[00:05:23] Jake: They're like, "Okay, this, you know, [00:05:25] god robot does this now, I don't have to think about it anymore." really the wrong way to use it. And you see people who are very effective with AI typically use it to expand on their knowledge and skills, and a lot of people that are not effective with it and really annoy their, you know, customers or their coworkers or anyone else that has to deal with their, you know, AI slop output, it's just kinda, you know, discarding the thinking part

[00:05:48] Nick: Yeah. Yep, and that's a

[00:05:49] Nick: [00:05:50] problem in any area of

[00:05:51] Nick: life. my philosophy anyway, right? It's like don't outsource your, agency part of it

[00:05:56] Jake: Exactly. And, and that's one of the, you know, one of the things, you know, we're kinda known for is it's, you know, sort of counterintuitive, but a lot of times, you know, at our core, like we're an AI consultancy. and so a lot of times it's like we start by telling clients like, "Slow down. You don't need AI here.

[00:06:09] Jake: You shouldn't use AI there." Whereas a lot of other, you know, companies that are just trying to sell you a tool will just be [00:06:15] like, "Oh, it's a great idea. This will work perfectly for you. Just go ahead and use it. Put it everywhere." And we're always like, " Whoa, let's pump the brakes on this and use it in the areas that actually make sense and use it the right way."

[00:06:24] Jake: Because it's like we want it to actually, a- achieve a, a desirable outcome at a point that makes sense and we're not pawning off thinking. So that's, that's one of the big flags that we always see. If you're using it to replace your thinking, there's no process, and someone's like, "I just want it to get me [00:06:40] more leads."

[00:06:40] Jake: And no process, no thinking, no strategy. It's just like, "I just wanna pawn it off to AI." That's just the wrong way to use it for

[00:06:47] Nick: I read a social media post that somebody said that they were using it for this, and that's what I want to happen for me

[00:06:53] Nick: too

[00:06:53] Jake: Point me to it."

[00:06:54] Jake: Right. Yeah, and that's what I always try to-- That's why it's like I like to just keep, give people like the blunt, raw advice of just like, you know, there's so much hype on the internet, and AI is still such a new thing, and a [00:07:05] lot of people that are wanting to use it, like, you know, a lot of people that are, like, running these companies are, are much, much older than myself.

[00:07:10] Jake: So it's like they haven't been, they're not, maybe not as technical all the time, and they haven't been around it as much. So like when they hear a claim like that, it actually seems true. And a lot of times, you know, people think, "Oh, the AI, it learns on its own, it gets better." Or, you know, I saw an article about someone said that it made them an extra million dollars this year, and it's kinda hard to differentiate 'cause, like, on [00:07:30] one side, there is a lot of really amazing stuff that you can do with AI that actually is quite mind-blowing.

[00:07:34] Jake: Um, but then on the other, everything's so blown out of proportion, and you can never tell, like, is this person posting because it's a brand deal, or did they actually do it? And I would say most of the time, 90, 95% of the time, it's a brand deal. This is, this is them making their, their paycheck. This is them making money.

[00:07:50] Jake: So, uh, you really gotta take all of it with a grain of salt. And that's, a lot of times when we're, [00:07:55] uh, starting with new clients, it's like that's kind of what we're dealing with is we're debunking all the rumors. Like, here's what it can actually do. Here, this stuff was probably stretched a little bit far.

[00:08:03] Nick: So w- tell me about whatever you're able to share about what, what's one of the most expensive or the most expensive, AI mistake that you've seen someone make so far, that you've seen firsthand?

[00:08:13] Jake: Yeah. So one of the ones that comes to mind, their-- the company, a, a lot of them are older, so then there's this one guy who's, you [00:08:20] know, considered the young guy, but, you know, even older than I am. It's just that, you know, he had used ChatGPT sometimes and made some, you know, cool stuff that he showed around the office like, "Ah, this one replies to emails, this one, you know, parses this document," and whatnot.

[00:08:32] Jake: So they wanted him to create then, they're like, "Okay. Well, you're the AI guy now." He just got dubbed that, never really wanted to So now you're the AI guy. So what he built then, the first thing that he tried to tackle was this, like, [00:08:45] AI document processor, essentially, which is, you know, kind of vague, but it's like to save, set a whole bunch of these company documents, uh, all in Dropbox.

[00:08:52] Jake: They needed to be sort of, like, parsed, the data extracted from them and put in either, like, their ERP or their CRM, and then have, you know, some, uh, boards checked off in ClickUp and whatnot. So pretty simple, straightforward project. So he just kind of whipped it up with Claude and, you know, over the weekend [00:09:10] goes ahead and deploys it, and it actually did, you know, kinda work.

[00:09:13] Jake: but then what they didn't notice is at the end of, like, it was between, like, one and two weeks, they noticed that the AI spend for that, because it was the only thing he had built so far, uh, it was, like, over $2,000. And so they're like, "Okay. We gotta pump the brakes on this.

[00:09:27] Jake: This is not worth that amount of spend." So then it's like they're asking him to fix it and add more stuff to it and whatnot, and he's just [00:09:35] like, " I don't wanna be a developer. Like, I didn't sign up for this. This isn't even my job role. You guys, you know, nominated me for this." So- Then we come in, they had mentioned, they're like, "Yeah, we tried to use AI, it's just way too expensive."

[00:09:47] Jake: And they explained it, and I was like, "That is not that expen-" I was like, "That should definitely not be that expensive. Something went terribly wrong." So digging into it, you know, talked to the guy, and we're like, "Hey, did you, like, log all the API, API requests and responses and [00:10:00] whatnot?" And he's just like, "What's that?"

[00:10:01] Jake: And " Okay. right, cool. So let me go in there and check this out." So we go in and look, and so essentially what it was is that there were a whole bunch of, it's parsing all these different kinds of documents. Some are mostly text-based, some are more like stylized visual, DOCX/PDF, whatever.

[00:10:18] Jake: So what the... Well, I would say him, but let's say Claude did, is essentially it's like it's, its processing part was [00:10:25] actually extracting all the text and then converting all the pages to images and then comparing them so it can see the visual elements where things lay out. There were some images in them too, and also has the text.

[00:10:35] Jake: And some of these documents were, like, 50 to 80 pages long. So it's doing that on every single page. And when we looked into it, we're like, "Well, what, what information do you actually need here?" And they're like, "Oh, like, the, the client's name, these few details, their address," whatever. And most of [00:10:50] that fell on the first page, and in rare cases, the second or third page.

[00:10:53] Jake: So it's parsing twice. It's reading all the images and the text for 50 to 80 pages, every single document. The highest, most expensive model with max thinking, which is completely unnecessary to process through all of these. So it's like, no wonder that you burned thousands of dollars on this to process some documents because it's just like you're going way overkill here, and you really need to scope it to, like, think out the [00:11:15] process, and that's what a lot of people skip.

[00:11:16] Jake: And not that this guy necessarily was trying to pawn it off to AI, it just, you know, he got nominated to do this, and he went ahead and did it. And then the real kicker was that, so everything was in a Dropbox. So it's just like, okay, simple. Just connect the AI to Dropbox, you process his stuff. But what he didn't really think of or mention is, like, oh, there's, like, years and years and years of documents in here.

[00:11:36] Jake: Don't just start processing everything. Didn't really tell it [00:11:40] that. So he starts it up, it's working fine, and it's just, you know, processing every single document in their Dropbox back years and years and years until he turned it off. So that's... Oh, and then another thing on top of that is that it's all, you know, it's all in this unsecured database that pretty much anyone can access, so that's, you know, a little bit more technical when you add, like, row-level security and all that sort of stuff.

[00:12:01] Jake: Wouldn't really expect him to know it, but it's like those are the [00:12:05] things... A- and I feel like what this really shows is just like- There's a lot of this and a lot of people ask like, "Can you do this yourself? Can you build this yourself?" And it's just like, yes, you could do about anything yourself. You could do your own sales and your marketing, and you could do your taxes, you could fix your car.

[00:12:18] Jake: But then it really comes down to like, is it worth your time, and what are the things that you're then not considering that could go wrong? 'Cause like I said, at this point it's not really can AI do this, c- you know, could it [00:12:30] handle this? Is it smart enough? It's, you know, about as smart with a computer as you or I am.

[00:12:34] Jake: You know, it could do actually quite a bit. Uh, it's just more about like, how do you deploy it the right way? Because just letting it go off and just like do anything is definitely, you know, you're setting yourself up for failure.

[00:12:44] Nick: Yeah, absolutely. So good point about the i- you know, is it worth the time that you're spending on it? I think it's good... It's been good for me, and I think it's probably good for a lot of people to, [00:12:55] uh, mess with it just to kinda see what possibilities are. It's good for your mind, uh, in doses, I guess, to, uh, help you explore, use as a thought partner, all those things.

[00:13:05] Nick: But I also think from first-hand experience that, it can make you dangerous enough to do some real damage. I could do something very similar to what that guy did, You know, I

[00:13:15] Nick: guess all it really takes is one setting or one permission that you don't understand, that [00:13:20] you've just given it like open access to something or unrestrained, access where, how are you gonna find out?

[00:13:26] Nick: If they, if they ... Someone hadn't seen that bill for two thousand dollars, like how long would that have gone on being just horribly inefficient

[00:13:33] Jake: Exactly. Exactly. No, it could have run for months, You know, you always hear the headlines that people love to quote, like the, you know, the Uber burned their whole AI spend in four months or something like to be fair, I don't know exactly what [00:13:45] Uber was doing.

[00:13:45] Jake: It's probably a lot of development work code, stuff like that, which does get very token-heavy. But I think people kinda take that, like, the headlines that always sell are, like, the fear-mongering ones, which is like, "Anthropic's new model is so dangerous it's not gonna be able to be released," or, you know, this, "The AI is getting to s- you know, it's spending too much, and it's never gonna work out."

[00:14:05] Jake: But it's like, when really look at it, y- you know, it's like these are kinda... I'm not saying they're one-off stories. [00:14:10] They definitely could happen, but if you do things the right way, again, not saying that any company necessarily did it wrong or right, but if you're applying it to, like, if you're not Uber or Fortune 500, you know you're kind of in a different playing field.

[00:14:21] Jake: And if you do apply it the right way, realistically, it's like AI should be, like, less than a 10th at most of the cost of a human doing something. Like, I have never, ever had built something, uh, for a client where they had to debate, [00:14:35] like, "Is this spend actually worth it?" What usually actually ends up happening is that they forget to refill their accounts top it off with credits because they put a couple hundred dollars in, and it lasts so long that they just don't even think to, like, make it a task, and then they're like, "Oh, it's not working."

[00:14:50] Jake: And I'm like, "Oh, you gotta, you know, put on auto refill, top it off a bit." But they're like, "Oh, that was, like, three months ago." And I'm like, "Yeah, it's y- you know, usually a task is, like, a fraction of a cent, so [00:15:00] like, .001, you know, cents." uh, it, it's actually quite efficient, and it's like you'd have to be doing some heavy work for it to, like, really exceed, a reasonable spend.

[00:15:12] Jake: And then also, I think people sometimes fail to, like, look at the outcome of, like, what was the amount of work? Like, if you had AI do $20 of work, are you really saying that was more than $20 of work a human would do? 'Cause I would argue it's probably maybe 10 times [00:15:25] more.

[00:15:25] Nick: Right. it's like kind of turned our perspective on the return of the economics behind labor, uh, on its neck, and it's probably gonna get worse. Like, that's just... I think the example that you just gave is really good. If done well, you can get a lot done for $20 in, tokens.

[00:15:45] Nick: You can also spend $20 on buying the time of a human and get [00:15:50] shit.

[00:15:51] Nick: Yeah. Yeah

[00:15:52] Nick: if it's running efficiently, if it's structured properly, built properly, then, $200 is not necessarily more expensive. It's, than 20. It's, it's more of an investment if you're just getting more of the, production that you're, that it was set up for is.

[00:16:10] Nick: But if you're setting something up haphazardly or [00:16:15] uninformed, then spending $200 does not necessarily get you a better thing than the cost going up does not guarantee that you're getting a better outcome, and having it do all the things, it doesn't make it better either, right?

[00:16:27] Jake: So what we always say is just like, if you-- if there's no solid process behind it, and this is where lot of, uh, AI projects go wrong before you even started because you don't have the right data for it. You don't actually know how the process work. It's how one person [00:16:40] thinks that the process works, and then five people working under them that actually do the work do it five different ways.

[00:16:45] Jake: And so they're like, "Well, theoretically, it should work like this." And when you automate something and you don't really know what you're doing, it's just like you just get automated slop. So I've never seen a well set-up system overspend a human or even come close, but I've seen a lot of very poor systems that just needlessly spend where it's just like, write an email back to a customer, and [00:17:05] for some reason it's reading eighty-four company documents to have all the context, and just like a ton of tokens every single turn, no prompt caching, nothing like that.

[00:17:13] Jake: so I could definitely see how you get those runaway costs when it's just unoptimized there. But, um, when you deploy it the right way, and that's-- it's kinda one of the things of like you don't know what you don't know, and you, you could do anything yourself. You can look it up online, and I think AI is really good at, um, sorta gassing people up [00:17:30] and just, you know, summarizing all of the hard parts.

[00:17:32] Jake: If I say, "Hey, how do I make the next SpaceX?" It's gonna give me like a five-step process, and I, I promise you it's not that easy. So just really over summarizes, and it's not gonna tell me the things that I don't know. Sometimes you gotta directly ask. So I always encourage people like definitely like, you know, definitely take the time to learn AI, spend some time on it, try to build some stuff.

[00:17:51] Jake: It's just that, you know, the clients that we're typically working [00:17:55] with are at this level where it's like they have a very important workflow, very sensitive data or important data, and they just can't risk, uh, someone on their team that is not a developer just going in there and winging it and leading this AI initiative that they don't even wanna lead in the first that's, uh, also another funny counterintuitive thing that we see a lot of the times it's like, you know, the, the whole point is to have people wear fewer hats and actually work in the [00:18:20] job that they were hired for rather than doing like, repetitive data entry into five different systems.

[00:18:24] Jake: That's-- It's just, you know, kinda silly to be paying someone a salary for that.

[00:18:28] Jake: and then it's like with this, sometimes they get more dumped on them, like the story I was telling you. So then they're wearing more hats now, hats that they don't wanna be wearing, and you kinda just added to the problem rather, rather than, uh, resolving it.

[00:18:38] Nick: Yeah. So how does someone recognize, like, when they're ready for it? Or maybe alternatively, when, [00:18:45] when they're, when it's not for them, when they're not ready?

[00:18:47] Jake: Yeah. So, I would say a lot of times, and this is another interesting thing, you know, it's just happened as AI improved of there's not a lot of processes anymore. There used to be a lot more processes where it's just like, AI is gonna suck at this, like two, three years ago. It's just like you're not gonna get a great outcome.

[00:19:01] Jake: Now it can handle just so much stuff. so we have like a lot of rule of thumbs, uh, that we kinda go by is the-- one of the first ones is like, [00:19:10] this actually an automation issue or a process or behavioral issue like we were talking about before? Like, is, you know, is this something that should actually be automated, or is this just like, for example, it's just like, let's say, uh, you want an AI to put in all these estimates because, uh, your team isn't doing it, but then to actually confirm the estimate, they need to call the client back, and they just don't call the client back.

[00:19:31] Jake: Well, you can automate the first part, but really what you have is this process [00:19:35] behavioral issue that y- the team is not-- whoever is supposed to be in charge of that is just not doing it. So there's, you know, sometimes in those cases, like no amount of AI that's going to solve that. However, you fix that part, the AI actually makes sense.

[00:19:47] Jake: So I would say a lot of our rules of thumb are like, what do you have to fix before you actually go into it? another thing would be, can you teach this to a new employee? Let's just say someone, uh, let's just say really smart college-level intern. [00:20:00] You know, they might not know anything about your company now, but could you reasonably teach this to them or, you know, bring them up to speed on it?

[00:20:06] Jake: If so, who is the one that would teach them that? Or is it that there's five different people that do it, none of them were really trained a certain way. They all figured it out on their own. but again, that can be solved. A lot of their, you know, there's like the process alignment there, where it's just like you figure out, okay, like what, what is the actual goal?

[00:20:23] Jake: What are the steps? How is each person [00:20:25] doing it? Which ones align? Which ones don't align? What makes the most sense? Okay, we found, we found the happy path. Let's automate that. And sometimes that actually ends up working a lot better because when you do these things at the same time of like, uh, we call it like optimizing the process for AI, when you do that and then also automate it and get everyone then doing it the same, it kind of helps solve the second issue.

[00:20:46] Jake: Because like if the AI does it this way, we all agree this is the best [00:20:50] way to do it, every new person or every person currently doing it now follows this AI workflow or, you know, a, you know, AI-assisted whatever, uh, partially automated, fully automated workflow to actually get the, the same outcome consistently every time.

[00:21:04] Jake: So that's one of the things that just kinda happens to fall hand in hand there.

[00:21:07] Jake: and there's a few more. So another one is just like, do we actually have data on this? Like, do we have the, data that [00:21:15] we actually need for the AI to make decisions? Are there any SOPs? Is this all tribal knowledge? This is a lot of where these next two steps are where a lot of people get stuck because it might be like, okay, we know how to do the process, but it requires, like, this task requires that we, look up this document, pull this field from it, check this customer in the CRM, read the past three emails, so then we actually know where to move ahead.

[00:21:39] Jake: [00:21:40] Because a lot of stuff, especially if you want, like, an agent to do it, it's more dynamic. You know, that's, that's where it really gets to be super useful because if it's just like, uh, when this happens, makes it-- make this decision, that's kind of how AI used to work, and so it's, like, partially useful.

[00:21:54] Jake: But when you think about the things that you and I do on the day-to-day, there's a lot of just like, "Oh, well, I gotta go get this information. I just know how this works because of my experience, and I've done this a hundred thousand [00:22:05] times." And you don't really actually think of, like, writing it down. tracking the process, all of that sort of like, what data do I actually need?

[00:22:12] Jake: So I think that's where it gets into, kind of overwhelming for people. I think that they might get partially along, and they're like, technically, this AI kinda does what it should, but it's, like, doesn't really have the full data to be, like, informed of how to do it right every time. So then therefore, it doesn't really work, and they're like, "Oh, well, going to get all this data and collect it is [00:22:30] gonna be too hard.

[00:22:30] Jake: I don't know where it is. I don't know if this is documented." Or maybe it's just, you know, they're an employee on a team, and IT is not gonna let them have all this data to actually go and train a model. They're not getting this and feeding it to AI." then the next one where people get stuck a lot, and this is where, you'd be surprised that it's, like, sometimes the most, uh, challenging and time-consuming part of these AI implementations is working with other people on the team and [00:22:55] vendors, um, specifically vendors.

[00:22:57] Jake: Let's say that you might use maybe your tech stack maybe is like ten, fifteen different platforms just depending on what you're doing, sometimes even more than that. And so sometimes in that data, it's like you might need, uh, API access that's only granted when you communicate with the vendor.

[00:23:11] Jake: They might have data on the back end that they would give you, but they don't really make it easy to get, so you gotta coordinate with them. So kinda knowing the right questions to ask [00:23:20] there and going back and forth with them because some of them make it very difficult, uh, especially if they're doing something with AI.

[00:23:25] Jake: They really wanna trap you in the platform. So it really takes, like, a lot of the, uh, coordination there, the back and forth, getting the API access, any data that you need, reports, all that sort of stuff can be such a pain if it's not publicly available. And, a lot of times what we're dealing with is, like, these sort of, like, proprietary systems or like, [00:23:45] custom builds that people have been running for, like, the last twenty years that were developed in the nineties or, you know, they're-- been working with a software vendor from the nineties that haven't-- hasn't updated their software since the nineties It's like some guy just like cooked it up in his garage, and it's unbelievably hard to work with.

[00:24:00] Jake: And another one of those things you don't know what you don't know. So how would you know the questions to ask if you don't really know, you haven't done it? that can sometimes be s- like the biggest [00:24:10] blocker and most time-consuming part. Like building the AI from how many times I've done it, pretty streamlined, quite, you know, simple depending on the workflow, but it's always like the coordination with the team, the training, working with vendors and whatnot.

[00:24:22] Jake: So I'd say those two are specifically where we see most people get stuck because once they realize this is actually gonna be real work to go and track down this data or coordinate with this vendor, they're just kinda like, "Eh, it didn't really matter that much."

[00:24:34] Nick: I've [00:24:35] seen where you've mentioned a statistic about getting like thirty percent more out of, out of your team as a whole. And when you, when you talk about the things that you were just talking about, it makes sense, right?

[00:24:45] Nick: So it's, you're not just, getting the AI to do something that's more efficient, but it has an effect on the people who are using it, and then even so far as like extending out to, how you're, you're, you work with the vendors, right? So there's a kind of a [00:25:00] cascading effect around it when done well

[00:25:02] Jake: it's kinda interesting because it's just like time savings, I feel like is the most obvious, uh, benefit of AI, and it's just kinda like any AI is gonna save you time, but also that's not like a huge strategic advantage.

[00:25:12] Jake: a lot of times is what we see is like it's sort of like the hidden opportunities, uh, the data and all that that you actually have, you know, somewhere. It might be locked behind a vendor, you might actually have access to it, and kind of like what you can do with that and [00:25:25] the insights that you can get from that.

[00:25:27] Jake: So it's just like, you know, getting thirty, forty percent more out of a team is just like kind-- I wanna say kind of a given at this point if a workflow is worth automating. They're kinda like boring use cases too. A lot of people look for like the most hype ones of like, "How will AI close me ten more deals this month?"

[00:25:42] Jake: But like a lot of times you wouldn't realize that there is, you know, twenty-five employees, and half of their time is spent on things that aren't their [00:25:50] job title, and there's like half a million in payroll that you're spending on things that really just don't need humans to be done at all. And then to, you know, end up automating that would cost like less than five, ten percent of, you know, what you're spending every year.

[00:26:03] Jake: You pay it one time, the AI works forever, and, uh, at a fraction of the cost that the humans are doing it. And so then there you've just added, you know, a half million dollars to your revenue, uh, from things that you didn't [00:26:15] even know were going on. So I think that that's like kinda like knowing where to look and what to look for.

[00:26:19] Jake: It's like you can chase the shiny objects, like you said, the posts online of like, "I just deployed this AI and it got me fifteen more clients." And it's like, yeah, that didn't happen. It just didn't-- It,

[00:26:30] Jake: Stay off Twitter. Stay off

[00:26:31] Jake: Yeah. It can, but there's a lot more behind the scenes that would go into um, you know, than just deploying an AI.

[00:26:38] Jake: But yeah, think it's, [00:26:40] it's kinda like knowing where to look of like what, what data would be useful that you're not using. For example, we have this client, they have, uh, multiple or quite a few gym locations. And so if you've worked with gym software, there's like a few big ones for like the check-ins, kinda acts like as a CRM, the calendar.

[00:26:55] Jake: It's kinda like this all-in-one. But like there's, you know, like a couple that really dominate the space, and they suck. And it's just like they're, you know, they kept promising. So they had-- They're working with one [00:27:05] vendor for years, and they're like, "We need reporting. We need better reporting." And like, "Oh, you can get a CSV of all these figures."

[00:27:10] Jake: That it's like makes no sense. Maybe if you're an Excel wizard, you could make something of it. But again, these are not tech people. And they're like, "Can you just give us reporting? I wanna be able to see these stats. I wanna see which clients are gonna churn. I wanna see our new members. I wanna see all this."

[00:27:23] Jake: And they're like, "Oh, you can kinda see one or two things here." So they just would not do the reporting. That's another thing. A lot of times larger [00:27:30] vendors will not actually give you something, you know. They're not gonna build around you typically. then they move to another vendor because then they promise them.

[00:27:36] Jake: They're like, "Yeah, no, we have way better reporting. We're gonna give you all sorts of reports," whatever. Once they sell them, guess what? No reporting still. how did we get the reporting?

[00:27:44] Jake: Yeah. Yeah, exactly. And so they're just like, oh, you know, they're just extremely frustrated, and we had been doing some AI projects for them, and they're like, "Do you ever work with this type of thing?"

[00:27:52] Jake: They're just, just kind of venting about it. And I was like, "Oh, [00:27:55] yeah. I mean, can we get the data from them?" And he's like, "I don't know." So I'm like, "Oh, let's talk to them." So I just talked to him, and I was like, "How do we export all this? How do we get the data?" There actually was a way to do it. Um, and so we were able to build this whole, dashboard, like analytics dashboard with also like an AI agent inside of it.

[00:28:09] Jake: And so from that, I mean, and, and this is just something, you know, it's like, yes, like analytics dashboards have existed for a while. We added AI on top of it to kinda give them those, uh, quick like, uh, chat [00:28:20] insights where they could just say like, where did most of our clients come from this month?

[00:28:22] Jake: What should we be focusing on? And it kind of makes sense of the data. But even just from having that and building them their own custom dashboard, it's like they were able to find, I believe it was 327 clients right away that were like high churn risk from everything that we compared, and over 400 that had done, uh, personal training before but never renewed their contract.

[00:28:42] Jake: So that's just in the first month. And so then they can, they can [00:28:45] reach out to them, follow up with them, and that added 130 grand of revenue just in the first month. And this is a gym, by the way, so like that, that recurs. Like people keep renewing. They, you know, they stay usually for over a year.

[00:28:55] Nick: You're talking about $130,000 per month of revenue.

[00:28:59] Jake: Yes. Yeah. Yep. There were-- Yeah, there were so many clients and this is just a half

[00:29:05] Jake: bat. Yeah, this is just right off the bat, just from being able to see those [00:29:10] insights, um, which, you know, we were able to put together, with them using AI as well to kind of see if just like a-have them ask the questions that they wanted answer, and then the AI can kinda go run all the queries and stuff and give them those sorts of answers.

[00:29:23] Jake: Um, so it works in two parts. Like you can see the stats and analytics. Also, the AI can just kinda tell you, like, "Hey, here's my read for the month. Here's what went well last week, last month." Um, so super powerful there. And again, it's one [00:29:35] of those things like, did it go out and find the five hundred more gym members?

[00:29:38] Jake: No. But like, they didn't really need that. It's just like they have many locations. They have a lot of members and stuff. It was just kind of like seeing, you know, making sense of the data that was already in front of them. So I feel lot of, uh, just knowing where to look exactly like that.

[00:29:51] Nick: they needed to be more informed. They knew what to do if they had the information, and so they used the tool to be more informed, make better [00:30:00] decisions. Yeah, that's a million and a half annualized revenue that found in the first month. that doesn't surprise me at all.

[00:30:08] Jake: Yeah, I would say, uh, probably a lot of companies somewhere in your data, like if you know the question to ask, which you probably do, it's just like, which, you know, what's the high churn risk? Where's our best opportunities here? You know, the clients that you have are your best sales opportunities right there.

[00:30:21] Jake: And a lot of times, you know, what we hear so many-- so much is just [00:30:25] people typically, like aren't doing the basics. It's just like, you know, is your sales team recording calls? No. Do they use the CRM well? No, not really. They just kinda put notes in when they feel like it. Sometimes they don't even put the deals in.

[00:30:36] Jake: They don't really move things along. So on the other end, it can actually help you. It's like, you know, even if you don't have the data, so let's say here it's like a, a sort of a data issue where it's just like you technically have access to the data. If you would like record this, what are the common [00:30:50] objections?

[00:30:50] Jake: Where are people going wrong? What's not getting added to the CRM? and you can actually build on top of that to then have agents. So it's like, what are the most common things? And I recommend everyone do this. This is not even something that we build, but everyone should have like an AI note taker, especially if it's like you're running a sales team or something like that.

[00:31:06] Jake: Like, why would you not record the calls? Like that is so, so much valuable data. And it's like, just go with like something like Fireflies. It's like 20 bucks a month or

[00:31:14] Nick: [00:31:15] Yeah, $15 a month and it, it'll do it all, everything. You just turn it on

[00:31:19] Jake: it's such a no-brainer. And then you, you can take it a step further there. And so a lot of times what we do is then connect that to like a CRM agent with... You know, you can put it into like HubSpot, Salesforce, Pipedrive, like whatever you want. and the cool thing about it is like, you know, and people argue like, "Oh, well, there's like Agentforce and stuff like that."

[00:31:35] Jake: But again, when you're using like vendor tools and stuff, they try to like lock you in there. [00:31:40] Like Salesforce doesn't really have, uh, an obligation or really want to help you, say, manage things be-between ClickUp and Salesforce or whatever. I mean, I'm sure there's kind of a way, but you know what I'm saying.

[00:31:51] Jake: Like, they typically like to keep you in their platform. So when you use-- When you plug into your own like ChatGPT or Claude account and give it access to all the different things that it needs to have, it's like you can connect it to Fireflies and your CRM and your project [00:32:05] management. And so then it's like one agent here is helping all of these different teams coordinate, and then immediately it's like you're getting way better data on every call.

[00:32:13] Jake: You're logging that to these actual contacts and deals so that someone else can then go in and see like, "Hey, what's the overview on this? Okay, we had all these calls. Here's the summary. Here's the main points." a lot of times we use it for like qualification too, because it's like that really gives you a lot of insight on why, a deal might go through or it might not.

[00:32:29] Jake: If you can [00:32:30] identify the, uh, BANT, we call it, which is a common one, just budget, authority, need, and timeline. A lot of times what we find is like people are sort of, uh, they're wasting their time talking to people that don't have one of those. And so it's like that deal was never really going to go through anyways.

[00:32:43] Jake: like to log things like that that they might not think off the top of their head of like, "I should write this down in the CRM right after. Did this person, illustrate their budget, authority, need and need a timeline?" A lot of people just, they wouldn't do [00:32:55] that. But on pretty much any call you listen to, especially like the first call, it's like you can identify that.

[00:32:59] Jake: You can put it in the CRM. You can then sort of like use that to analyze and rank the deal and see, how valuable this might be. there's just so many use cases there, and kind of the point that I'm making is like you can use AI to actually give yourself that better data to do all this data collection and, analyze it to see what you could actually be doing.

[00:33:18] Jake: So then you can take that in [00:33:20] the case like the dashboard and actually make better decisions on it, and that's kind of where you see the improvements. So not everything is gonna be the shiny object right in your face, "Oh, it just made me a million dollars from nowhere." It's like, you know, a lot of times you gotta know the right questions to ask and know the places to look.

[00:33:35] Nick: Yeah. Well, I think sales is a g-a great example of where, the, you don't want the AI to come in and, do all of your job for you. What you really want [00:33:45] is you to be able to do the most valuable part and do a lot more of it and not worry about the rest, and the AI to take care of all the rest or as much of the rest as possible, and that's what enables you to be able to just focus on building the relationships, talking to the people, closing the deals.

[00:34:01] Nick: Like at this point, I mean, you, you may correct me if I'm wrong or you may feel otherwise, but I'm like, there's no reason for you to not [00:34:10] know, the pulse of your pipeline within minutes of real time. And I mean, if, if you want it, like you should know all their vitals, like for everything in your pipeline, if that's the depth that, that gives you value

[00:34:21] Jake: Absolutely. Yeah. The-- It's, it's 100% possible, and again, I, you know, like I said, people chase the shiny object, but it's like you, you illustrated it perfectly. To get 30, 40% more out of your sales team, they're all doing sales [00:34:35] admin work, and that usually takes, you know, that's an easy one, like 30, 40% of your time that is just wasted there.

[00:34:40] Jake: So it's just like, a somewhat obvious equation. If they had 30 to 40% of their time to then go and sell more, would you not, you know, potentially get about that much more business?

[00:34:50] Nick: Yeah. And you think about the, like your, your real killers in the sales department, like what are they gonna do with that?

[00:34:56] Jake: Exactly. A lot of, a lot of the times what we see that's, [00:35:00] um, missed and-- is, like, the referrals and follow-ups. It's like people just get so behind because of these sorts of tasks that we're "Oh, your referral process?" "Oh, we don't really have one. We kinda thought about it, but no one really has time to make these calls and follow up with clients after a month or two, see if they're happy, help them out if they're not, and then, you know, get referrals while they are really happy."

[00:35:20] Jake: And so then it's like another thing you really gotta consider, and this is, when you think about an AI [00:35:25] initiative as a whole, it's so much more than just, like, implement these tools. Like, everyone just use ChatGPT and we solve this. The next part would be once you've actually had a, you know, successful rollout, is what are you gonna backfill that time with?

[00:35:37] Jake: Because some people, you know, if they're real go-getters and stuff like that, especially, you know, a lot-- usually more, higher level salespeople because they're on commission, whatever, they're more self-motivated. They-they'll probably just apply that time better. Not always, though. A lot [00:35:50] of times you see people that it's like, "I have 30% more time to do nothing.

[00:35:53] Jake: I'll just kinda 'cause, like, the same amount of work gets done. So that's when it also comes into, like, the process behavioral change of just like, what are we actually gonna do with this time that we've gotten back? Like, what's the next, uh, high-level initiative? And what were you actually hired to do?

[00:36:08] Jake: Because I can't think of any client that we've worked with where everyone on their team was doing their actual job title. It's the [00:36:15] opposite. They're all, you know, wearing 20 different hats and they're like, "This is nothing like what I was hired for. I don't even... I can't even describe what I do, like, what my job actually is."

[00:36:23] Jake: It's like j-firefighting, essentially. that's another thing, and I feel like that has to come, you know, from leadership. Like, I can't really dictate to people, like, "This is what you need to backfill your time with." But it's a good thing, for, leadership teams to think about of just, like, let's say in a perfect world, you eliminate all this work, you have these better insights.[00:36:40]

[00:36:40] Jake: W- you know, what are you trying to get? Like, what's the goal here? What insights would actually give you the highest ROI if they were acted on? And then who's going to act on them?

[00:36:48] Nick: Mm-hmm. So, I mean, there's just, it's like there's a lot of education or informing that needs to be done, seems like at every step of the way, which, you know, that's not out of left field. Like we, I think we were talking about before we started, I call it the Wild West. man, it's changing fast. So [00:37:05] it makes sense we need people like you to be able to, keep us in the guardrails, maybe keep us safe. so what does it look like for somebody who is just, they're interested, trying to figure out how do they need to le- be leveraging AI?

[00:37:19] Nick: if I point them to you, what's your advice or where do you direct them?

[00:37:23] Jake: Yeah. So I actually just built recently this free tool for anyone to go use. It's an AI opportunity finder where you can sort of bounce [00:37:30] your ideas against it, and it tells you-- It kinda thinks a little bit outside the box. 'Cause you can go, like I said, I can go to ChatGPT, how do I make the next SpaceX?

[00:37:38] Jake: It'll probably give me a five-bulleted list of the

[00:37:40] Jake: You're going to get an answer.

[00:37:42] Jake: yes, yes, and it will, you know, somewhat kind of be the right one. But, uh, you know, it can be such a yes man sometimes. So what this does is actually breaks it down, thinks outside the box of like, here's the things to consider, of just kinda the things that [00:37:55] I ran you through of like, where does the data come from?

[00:37:57] Jake: Could you teach this? We need some more info on it. What are the exact steps? Do you know these are the exact steps? It really walks you through all those questions that you really need to be answering, and then gives you this full report of just like, here's how this would be automated. And then also on top of that, which I would argue is a little bit more useful, is these are, are the things that you should work on in the meantime, that you need to kinda prep this to be AI optimized so you can actually go ahead and [00:38:20] automate it.

[00:38:20] Jake: So, I'll send you over a link to that. It's free. There's no sign-up. You can have it, the report emailed to you if you want, but you definitely don't have to. and yeah, it should give you some really good ideas of where to start

[00:38:30] Nick: Yeah, so I have this open right now. So this is, it looks like, it looks like a quiz, but there, I'm actually communicating with bot or an agent here, right?

[00:38:41] Jake: Yep. So it, it is an agent behind the scenes and, uh, so it's dynamically [00:38:45] generated. It's not just like a set answer thing. It's like you'll start to realize that it's like the questions are generated off your answers. So it's just a little bit more, um... keeps it kind of more on the rails than if you just have a chat, window.

[00:38:57] Jake: So it kinda guides you along, you know, asks the right questions, gives you some ideas on how to answer because that's what I found is, um, a lot of times people, it's like they might have the answer in their head but they just-- If you ask them to type it out, they over summarize it. So I highly [00:39:10] recommend there is a, uh, uh, a speech-to-text button on there, a little microphone.

[00:39:14] Jake: Uh, and I would recommend this as a huge tip for using any sort of AI talk to it. Like, just use like ChatGPT and Claude habit. Like A lot of times people don't realize how much they summarize something and that's why they get a really crap answer. Because they have the information in their head but they didn't really realize they didn't communicate it.

[00:39:30] Jake: So like I, I recommend for each question like i- you know give it like, [00:39:35] talk to it for like thirty seconds to a minute. Give it all the context that you can because it's just like that's kinda how you, uh, help yourself in this sorta situation. If you summarize it to like, "Yeah sometimes I send out some emails and these documents take a while," it's hard to really give you a really good actionable report of what to do next so

[00:39:51] Nick: Yeah, and I think for, for me, I've noticed when I've been doing that, and I use w- voice-to-text for a lot now. for thi- [00:40:00] for ideas that are not, that are not, like, fully ironed out, there are elements to it that, are probably, I have them buried, that end up being, like, material information that it needs to know.

[00:40:11] Nick: And, so either the process, like, of me getting it out, I realize that, or working through the process with AI, I'll see that it factored in, and I'm like, "Huh." It now makes sense after I've kinda gone through this [00:40:25] process. So, if you haven't f- already, like, filtered or sanded down your idea, that's when I really like using the, the chat, because you can just dump.

[00:40:34] Nick: But again, it's going to give you a- an answer. It may not be a good one. It might, you might think it's good, but it's, you're definitely gonna get an answer.

[00:40:42] Jake: Yeah. Always gotta with a grain of salt, but it's just like, you know, the more, the more you use it, that's again why it take some good judgment on the output [00:40:50] of

[00:40:50] Jake: It'll strengthen your discernment

[00:40:51] Jake: Exact discernment, that's the word I'm looking for. So it's like you gotta, yeah, you gotta really understand what's a good output and what's a bad output.

[00:40:58] Jake: But, um, another thing I was thinking, I don't know if you have any insight on this, but I know that, you know, you work with a whole bunch of clients, and the first time that we talked you seemed to have like noticed a lot of the same things that I did and have some pr- you know, specific insights on working with AI.

[00:41:11] Jake: What have you kind of been seeing out there in the field?

[00:41:13] Nick: Well, at the beginning of our [00:41:15] conversation, you, I think you described it well, where it's like, "Oh, this thing is gonna come and, and do all the work for you. Let's do it." And then we were-- a lot of people were overzealous with the, delegation or abdication, and, uh, now I feel like there's been a correction or we're in the correction where, you know, the people who want to think for themselves are like with snapping out of it.

[00:41:39] Nick: And, [00:41:40] uh, like, okay, this is kind of where we're at with it right now. And so they're having conversations like, you know, with people like you and taking a, a more, a more grounded approach to it. I think that there are still other people who are not interested in... Like, they don't want to have to think.

[00:41:58] Nick: they're kind of looking at it as th-this is gonna be the thing that replaces people or how can I replace myself? [00:42:05] people who are looking at it like that are gonna get replaced, it's not probably not gonna be the way, way that they think it is, but yeah, I think it's gonna happen.

[00:42:13] Nick: And then the people who still wanna retain their agency and, and want to exercise some discernment and, like, aren't trying to offload all of their, the value that they can provide This is a huge opportunity. I don't have any idea what direction it will go, you know, in a few years down the [00:42:30] road.

[00:42:30] Nick: But for right now, I think the people who who see it as an opportunity for leverage and to, to amplify the value they can create rather than to, let them do less and create the same amount of value, I think those people are, are gonna do really well. I think it's a great opportunity

[00:42:48] Jake: No, I think you really nailed it there of like people that make themselves replaceable will be the first ones to be replaced. So it's like there is definitely so [00:42:55] much fear years back of this thing's gonna come in and take my job next week, and uh, we've seen that hasn't necessarily, been the case.

[00:43:01] Jake: And a lot of times where we kind of see it 'cause it's like we always advise against a lot of people like, "Okay, so we'll have an agent now and then we can just fire all these people." It's like whoa, whoa, whoa. Let's, let's the brakes there. You know, you still wanna have the people involved, make sure it's doing the right things and these are the, people that are actually going to like think outside the box and, you know, actually [00:43:20] make these processes better and have better ideas for what should be automated and whatnot.

[00:43:24] Jake: Like they're the boots on the ground type of people. But what we actually do see a lot of is when, companies are like, "Oh, we've outgrown our process and we're gonna need to start hiring soon." And then a lot of times when they can... You know, that means that they have that excess of work to sort of backfill when people, uh, have that time freed up and I feel like that's a, a much better way to go, go about it.

[00:43:43] Jake: But definitely the people that use it [00:43:45] to not think, it's like this has probably been a trend with you for a while. You know, you've probably been trying to cut corners and whatnot. You're maybe not the biggest team player or biggest contributor so you might have been replaced anyways but you're right, that's a-- it's a surefire way to make sure that you're the first one to be replaced.

[00:44:00] Jake: But with your, you know, like companies you work with, have you seen any that you feel like have been really successful in rolling out the-- an AI initiative? Where you're actually [00:44:10] like, oh, that's impressive, like they actually did this properly. There weren't horror stories. They actually, you know got the equivalent of many full-time roles of work out of different AI tools and agents

[00:44:20] Nick: I don't think I've seen anything that I would call a catastrophe, firsthand yet. probably the thing that's gotten my attention is just the different, the creative ways that people come up with to apply it and solve sometimes small [00:44:35] things, sometimes like bigger issues for themselves.

[00:44:38] Nick: I mean, Like smart people are, are coming up with like clever ways to use AI for themselves. And I mean, Like some of it's as simple as a, coaching business and them being able to like standardize and, codify all of their IP and build training out for staff and like handle a lot of the administration.

[00:44:59] Nick: [00:45:00] So, they were able to add on capabilities for onboarding people that they didn't have before. And so, they didn't have staff that were in charge of that, that they got rid of, but that gave them, like rails. they're able to bring new coaches on, get them up to speed, and, and can do it just about as fast as they want to.

[00:45:21] Nick: Much faster than they, can fill a coach's [00:45:25] book of business for them. you know, that's pretty powerful, because now that's on to the next thing. Now they can focus on being able to build that coach, the coach's book of business and, you know, then they can maintain the capacity. And I've seen agencies that, that are doing that.

[00:45:37] Nick: like what we're doing is, building out like content infrastructure one of the thing that with the AI slop, a lot of people have, try to use AI to create content. [00:45:50] and what we've been focused on is the infrastructure behind the content, so it contains the, the elements that it needs to contain or that you want it to contain about your brand, so your brand is represented through all of your content coherently and consistently.

[00:46:08] Nick: And it's like that is just something that we knew before AI was an issue. The use cases just keep piling up, but [00:46:15] it's just people figure, figuring out, getting over the fear, probably. Getting some confidence through like, "Hey, I'm gonna test this." And then we, give them a horror story about somebody's getting loose and spending two grand in tokens.

[00:46:25] Nick: But, know. But, but two grand is like that's not the end of the world, right? If so, if you spend a few extra hundred dollars to learn something, you can consider it an investment in your education because you're gonna, you will [00:46:40] be educated by this at some point. That's my opinion

[00:46:43] Jake: 100%. Yeah, I don't mean the, uh, you know, the horror stories to, like, scare someone away from it. I think people should definitely use it, experiment with it, and it's exactly like you've listed one of the best, like, personally use cases anyone can get started with is just like organizing, mapping processes, getting data together.

[00:46:59] Jake: So it's like you could actually, you know, it can help you with those, like, prerequisite steps to actually automating [00:47:05] something. But yeah, specific example I was talking about, that's kind of when you get into like, you know, building your own app that handles this entire workflow and plugs into the API.

[00:47:13] Jake: So, um, if you keep it in your ChatGPT or Claude window, you definitely, uh, minimize those, minimize those risks for sure

[00:47:20] Nick: Yeah. I mean, but it's here, right? Like it's, it's not gonna go away. So there's-- I don't know that it does anybody in the workforce, uh, or [00:47:30] in businesses any good to run from it

[00:47:32] Jake: Yeah. It's one of those things, I mean, I know a lot of people, especially when it comes to, like, the AI art in movies and stuff, they're just preying on its downfall, and they're like, "Oh, it'll fizzle out." But it's just like, you know, humans have always been resistant to change. I don't going anywhere.

[00:47:46] Jake: You know, there's all the people, you know, "Oh, it's a bubble, it's a bubble." It might be, but I, you know, I, I think it's just-- it's gonna survive. it used to be like, "Oh, you gotta catch up. Your competitors are [00:47:55] using it. Just, just get AI everything, and it's gonna be so great."

[00:47:58] Jake: And then people are feeling kinda like the downside of that and, like, where it can go wrong. Because so new, you don't know what you don't know. You don't really know where it could go wrong. It just seems like a good idea, like how bad could it be? And then you see how bad it could be.

[00:48:11] Jake: And so I feel like now we're in the phase of people actually being like, "Okay, I need to be intentional about this. I need to, you know, actually implement this in a way that's going to [00:48:20] work." Just, throwing shit to the wall and seeing what sticks is, is not how you actually accomplish this. AI is-- it's not gonna do it itself.

[00:48:26] Jake: You don't deploy one agent, and it just keeps getting smarter until it knows your whole company and replaces you. Like, that's another funny thing I think about all the, uh, clickbait stuff online. It's like just ask yourself when you read something online, "If this existed, wouldn't everyone be using it?"

[00:48:40] Jake: Like, if it's really a, you know, an agent that made someone a million dollars in a [00:48:45] week, and it costs them $20 per month, don't you think everyone would be using that, and you'd see it everywhere, and everyone you knew would be using it, and we would

[00:48:51] Nick: Everyone around the planet would know within a day, right? If it was that good, ev- within a day everyone would have it

[00:48:57] Nick: it

[00:48:58] Jake: Yep, yep. Every, and, and it's one of those, it's just like, uh, what was the movie it was from? Maybe The Incredibles or something like that, but it's like something along the lines of if, if everyone has superpowers, no one has superpowers. [00:49:10] not where the quote originated, but that's where I remember it from.

[00:49:12] Jake: It's kind of the same thing of just like, yeah, if everyone has this, you know, infinite god AI that, you know, knows more than, you know, the smartest person on earth and can get you a million clients every month, then, we would kind of all be, we're kind of at a level playing field would certainly know about it

[00:49:28] Nick: Right, and we would have different problems to worry about at that point. So when, when it happens, everybody will [00:49:35] know.

[00:49:35] Nick: Exactly

[00:49:36] Nick: So tell me about what you're doing with, uh, Agenic Brain.

[00:49:40] Jake: Sure. So Agentic Brain is a-- at, at our core, we're really an AI consultancy. I started it with my co-founder, Frank. He's been a sales and a business consultant for over seventeen years, and then I'm the, you know, AI technical lead. And so kind of the problem that we saw with a lot of the, you know, AI companies and tools and stuff, [00:50:00] like I explained, they're, they're just trying to get a sale.

[00:50:02] Jake: They're gonna promise you the world, say it fits for any process. And when you apply AI that way, just kinda slapping it on, it's built around a general process, someone else's process, the results are always so limited. so then what we've noticed that really worked is actually taking the, um, consulting approach to it, talking to a company, talking to all the team leads, talking to the people that actually do the work.

[00:50:22] Jake: And this is typically where you find those gaps and [00:50:25] opportunities of like what someone thinks happens, and then this is what really happens. Going in there, helping them with the, um, first steps of actually getting the process AI optimized, then building AI on top of a working, working, proven process. And kinda how we describe it is it's more like we're building AI employees here.

[00:50:44] Jake: You know, everything has agents now. You open up Gmail, it has an agent. You know, Slack has an agent, you [00:50:50] know, all this sort of stuff, and it's like it's kind of lost its meaning. So what we found to work best is when you build actual, and train AI employees on this proven process and then go and deploy them, The adoption is so much better. Like, people go ahead, and they just start using it because there's, like, no learning curve for them. There's no blockers. So kinda two parts, understanding the exact challenges, needs, and goals, and [00:51:15] then doing intentional AI implementation and actually seeing it through to the end.

[00:51:20] Nick: Mm-hmm. Yeah. So there are a couple things that you-- I've heard you say that I think are really cool. One of them is, deliberately anti-hype, which I think comes through in the conversation as not boring, but, keeping it real, like somebody that you can trust, right? You're, you're not just gonna cherry-pick the information you're sharing.

[00:51:38] Nick: The outsourced [00:51:40] AI department, I think that's really cool, and listening to you talk about the being the consultant and how you've worked with some of your, clients and, and other companies to help them get their head around whether or not this was even something they need to do, I think that, that's, like, very valuable.

[00:51:54] Nick: And did I see that the AI employees, did I see that you guys have done, like, seven hundred, something like that, AI employees?

[00:52:01] Jake: Yeah, we've been doing this since, um... I actually had a-another company [00:52:05] before this that we merged into Agentic Brain. But since, uh, what was it? GPT 3 or GPT 3.5 came out and you could actually start using it for this stuff, I was already on it right then. And so, um, yeah, we've been lucky to work with, you know, a, a lot of different clients in over forty-two industries, I believe, a-all across the world, like eight different countries, uh, regulated industries like HIPAA, FINRA, GDPR, private [00:52:30] equity, gyms, uh, you know, like home improvement, you know, really all these different industries you can imagine.

[00:52:36] Jake: I really think any, any company will benefit from AI in some way. I've yet to talk to a client where I'm like, "Oh, there's no opportunity here. There's nothing you can do with AI." You know? We always find a lot. Yeah, you also brought up the, uh, outsourced AI department, I feel like is also kind of important because I, I just don't believe that AI [00:52:55] implementation is a one and done.

[00:52:56] Jake: It's not just you buy something, you forget about it, the AI somehow gets smarter and then, oh, we completed the, you know, AI initiative. We're, we're an AI company now. it's really this kind of like long term of like it's, it's a tool, and there's a lot of different workflows you could automate with it.

[00:53:12] Jake: So really where we see the most benefit of just like working with c-clients long term, and then it's like you, you just kind of [00:53:20] discover those things that it's like you wouldn't have initially thought to look for, just like, the vendors not giving the client all their, data or, or the reports in a way.

[00:53:27] Jake: And so it's like we work with them over a long period of time, and then you start discovering. You might additionally find ten workflows, but then once you automate those, you'd be surprised if you like talk with the team, the people actually doing the work, they have like a million ideas in their head of what automated.

[00:53:43] Jake: And once you hit those big first ones, [00:53:45] then it's like, "Oh, can we do this? Can we do this? Could it also do this?" And that's really where we see AI, becoming the most useful and getting improved the most when you sort of like build on it over time, 'cause no one has all the ideas upfront, and you don't know what the new things that you'll have time for and discover that you might wanna automate down the road

[00:54:00] Nick: Well, congratulations. I really like the perspective that you have, and, you know, the way that you're going about things. It's just... I mean, you can imagine being, the non-technical [00:54:10] person who i-- Like it's a breath of fresh air, right? It's such a relief to be able to, find resources that you can have conversations about these things with and be the dummy and trust the person that you're talking to because, I mean, you have told people that it's not for them, that they're not ready or not to do these things.

[00:54:26] Nick: You're not just coming up with a reason to sell someone something because you could. So I appreciate it. and thank you for taking the time and [00:54:35] sharing all this information. we'll make sure that we get the opportunity tool. You've given me a link for that. the AI Opportunity Finder Tool, find.agenticbrain.com. But we'll make sure that's all in the show notes, and we will, keep sharing the information with everybody.

[00:54:49] Jake: So

[00:54:49] Jake: Perfect.

[00:54:50] Jake: yeah, I appreciate you having me. It was a great conversation