Beyond the Demo: Why Most AI Implementations Fail and How to Build Ones That Actually Work

Speaker A: So look, first of all, uh, thanks for doing this and I appreciate the kind of collaboration we're working on. And you know, as I said, right. I, I have a lot of people that I speak to and I'd say let's call it like for the sake of it your space, which I call it the AI space specifically. Right. And most of the time I'd say you and one other person who I spoke to seem to be what I would call expert practice practitioners. Like, you've definitely cracked the code on figuring out ways that these tools can be used in ways that make a difference that I see is really significant. Right. Like, I've seen a lot of use cases where it's like a little thing like make this faster or even the stuff I use every day, but really being able to produce work product that moves things in a way that's really integrated. I don't see it often and in fact the instances I've seen you hear this probably more than me is it just doesn't fulfill on the promises that a lot of people are claiming. I see a lot of these weird things going on on the Internet. I built this bot that does this thing. It'll make you eggs and you know, take your wife out to D all in two minutes, you know, and it just doesn't really do that. And it seems like a lot of these things are built for people who are more engineers than just everyday users. So I would love to get your perspective on this. Like, what do you think really is going on right now in the state of AI with all combination of the hype and the reality from your perspective being I would call a practitioner.

Speaker B: Yeah. So, uh, let's start with the hype first because that'll uh. And then I'll, I'll get into the reality of what I. Cause I'll bring it full circle. But like there's so much fricking cash dumped into these AI companies. And I don't say dumped in a negative way. I just mean tons of EC money, tons of public cash basically put on a future bet. So just like what used to happen when software was kind of announced as vaporware, right. Where people would project what the opportunity was. I think there's an element of that, but not totally that. However, I think the market perceives it as that. And, and what I mean by that is like there's. And uh, I've seen companies started to course correct on it. When I say companies, I'm talking about the big ones like you know, Google or Anthropic or OpenAI. They, in Microsoft, especially in Salesforce, I guess those are kind of the, the ones that come to mind, um, and even Meta. But what they're doing is they're basically saying this tool is going to, became available and they're showing stuff that's six months out, nine months out, 12 months out. So they'll talk about it, they'll do a big PR push and then nothing happens because nobody could use it. Or they try and it's a fraction of what was shown in the demo because the demo was sped up or there was that incident where that happened with Google. Right. So I think those are a couple of things of why the market's a little bit jaded. However, at the same time, if you really truly understand how to use the tools, there are insane opportunities. And so I think there's a massive variation of skill sets in the market. And so the reality is like I've seen basic individual contributors and I've seen all the way to executives to CEOs and even investors basically go from being a noob to being an expert in like a month if they just have the time, focus and energy associated with it. And here's the number one thing I would say, Warren is like, because of AI, there's more tools than ever before that are available that uh, are coming out at a faster rate. So what happens is people are getting access to all these tools and if they don't see immediate time to value, then they're just kind of abandoning it. And I think that's happened for a lot of people. The other thing that I see is people are basically using AI tools like the large language models like ChatGPT or Gemini. They're using it like Google, like a Google search. Right. And that's really not what it's for. So that's, I hope that brings it full circle.

Speaker A: It helps a lot. And, and I, uh, it reflects my experience, right? I mean I, I, I'm both on both the delivery, not, I'm not building these things, but I'm, I'm, I'm a practitioner in that I'm helping people solve problems in the, in the realm that you are. And I also use this stuff. So I find too, I get impatient. You know, I, I see a video that looks really amazing. I had a really interesting conversation with an engineer about this about two months ago. So there was a product that came out. I won't mention it, I don't want to get into that, but it was cool. It was basically sort of like these, these guys created what ostensibly looked like, like Jarvis from Iron Man. You know, you can text it, you could talk to it, you can ask it to do stuff, and it does all these things. And, you know, the demo was really cool. Like, they showed this guy doing stuff during the day and talking to it and doing cool things. I thought that'd be cool. And it was so cheap. It was like 20 bucks a month, you know, So I figured, fine, I'll try it. So I downloaded it and it just didn't do anything that they said, you know. And so I got the engineer on the phone and they were really nice, you know, they jumped on and they said, yeah, you know, what happened was, it was a typical story. You know, we, we, we, we did this really great viral campaign, which, by the way, was really clever. Like, their marketing was really good. And so it was, it pulled me into their funnel and they said, what happened was we got so many people using it that it broke. Like, it just won't. We can't serve everybody. Like, it just doesn't work for all these people. And then the other thing they said was, oh, yeah, you know, that particular feature is still sort of. We're working out some kink. So what I said to them was what anybody would say. I was like, well, why'd this then? You know, I said, I don't get it. Like now I'm, um, not happy. I'm probably going to tell other people I'm not happy, you know, And I think the problem, uh, maybe you can correct me on this, is that there's got to get this thing out the door because there's so many other things going out the door. They better just get it out the door and fix it while it's flying than get it out when it works in the way that they promise. And so there's this weird devil's choice they have to make between getting it right and waiting or getting it out in the market and just hoping and praying. And I think that's where a lot of this stuff gets broken. Is that what you're seeing as well?

Speaker B: Yeah, I, I think I see some of that with, with companies. However, there's companies that I'm seeing ship really fast and do really well with it now, I think for some reason. And like, here's what I'll tell you. Like, from a big company perspective, I think Google is massively improved. Uh, when they first started off with Bard, I don't know if you remember,

Speaker A: I saw all that stuff. Yeah, it was crazy.

Speaker B: It was an, it was an abomination everybody's like, how did Google mess this up? And now, like with Google Labs, I'm also an alpha tester sometimes for different offerings that they have. And so I get early access and they're shipping like really fast. So I think they're putting a lot of pressure on OpenAI from that perspective. So that's one example. Anthropic's doing the same. Um, and then another company that, that I like is called genspark and, um, they're shipping really fast as well. So those are a couple that come to mind. But yeah, I agree with you, man. There's a lot of it where. And at the same time, by the way, the best feedback is like real feedback, you know, So I think part of it is like product development, like, they'll try it out a few times. And that's what's different about AI. It's not like a SaaS tool where it works every, the same exact way every single time. And if you don't have the right guardrails, it's going to be all over the place. So I bet you that's what's happening as well.

Speaker A: Yeah, I get it. And you look at a company like Google, I mean, they've built just, first of all, aside from the immense, we'll say, trust they have because they have just billions of users that they can make mistakes like this and fix them because they've just got this massive brand that they can sort of afford to do this sort of thing and they can recapture people back into their world again even if they screw things up. Mainly because, A, it's Google, right, and they have ridiculous resources and money and they're brilliant. The other part too is that, you know, I'm sucked in their network. I'm using other tools already anyway. I'm already in their world. I'm, I, I can't get out of Google, you know, so I got to stay in there. They have an advantage. But for, you know, incumbents that don't, they don't have that, they have to sort of like build that trust. And I think it's difficult. So where do you go to see this kind of playing out? Like, do you see it as being. And maybe you don't know, but you got a good perspective. Do you think it's like there's going to be, I don't know, five players really ultimately and that's it, or is there going to be a whole slew of library of different players out there? Is it going to be a kind of bundled offering that People are going to have or I log into something and there's a whole bunch of things that it does for me or is it going to be specialized? Or I've got a tool for this and a tool for that tool for this. Like what do you think the average user's kind of world is going to look like around this thing in like maybe a year or two?

Speaker B: Yeah, I think it's going to, I think it's going to mimic a pattern we see in normal SaaS and you either have a uh, point solution or you have like a do everything solution. But what I mean like that for large language models is like. Or AI models like and just for context like ChatGPT or Gemini is what I would consider basically a foundation model. And then a point solution might be something like Gong or. Right. Or Salesforce. And Salesforce, what doesn't want to be call the point solution but effectively they're a point solution. Right. Um, and so those are examples right where you could do that. Microsoft, um, Copilot for big companies is considered more like a platform solution. But what I would say is I think on the, the AI side of the house there's going to be like AI model aggregators that have like multi agentic function and then there's going to be just the core LLMs as well. Right. And then to take it even a couple steps further, there are going to be point native AI solutions that weren't created from a SAP solution with a bolt on AI product. They're going to be AI native. So you have AI native, you have a large language models and the obviously large language model agentic orchestrators if you will. Um, from that perspective. So that's kind of like the stair step that I would say.

Speaker A: Got it, got it. I'm hearing this word orchestrate a lot. You know, there's automate, there's orchestrate, there's agents. Could you give a little distinction of what these words mean for the layman?

Speaker B: Yeah man, it's a good question. Um, I think there's like kind of three stages when you look at it and I think it's funny because everybody wants to pass number one and it's the easiest to execute on. It'll get you the fastest roi. That's compounding over time with your largest expense and that's, that's augmentation of your existing team. Right. Because like I'm seeing teams that. And so I would say step one is like augmentation. So like that's evolving everything from AI, ah, literacy to being able to tactically use the use cases and models to deliver a business outcome. And so that's like step one, right. In my opinion, because AI workflows are top down and bottom up, you need to understand both angles. And without having that foundation in place, if you try and move to stage two, which is more like automation or agents. Right, Automation or agentic automation, if you will, that's effectively going through let's say a multi step process where you give it the goal or outcome and then it delivers a result. Okay, so that's number two.

Speaker A: What might be an example of. I'm sorry, if you don't mind. Be good. Just real world example. What would the example of an augmentation be in everyone's world? Like what's one that they know about that you'd consider that category?

Speaker B: Yeah. An augmentation would be like let's bring it back to sales. Right. Cause I know we got a lot of sales listening. An augmentation would be. And I've seen this and I've deployed this with companies effectively. We create like a deal strategy coach for a complex sales process. So effectively after every call transcript is fed into that, that custom GPT and that custom GPT basically scores the call, um, with whatever sales methodology are you using, let's say medpic, uh, and then identifies like a grading and also like what the person did well, what they didn't do well and need to work on. But then also next best action. Okay, so that's an example. An augmentation can range from like long form projects where it might, you might compress a uh, three hour project into uh, three, you know, to ten minutes. Right. Like that's an example. Or it could be the death by a thousand paper cuts. So that's like m. Let's say for a rep might be for account research. Right. Where it's not hours, but it might take a half hour if you do it right. And you can trim it down to three minutes multiple times a day, multiple times a week. That stacks really fast. Um, and that multiplies the capacity of the team instantly. That's, that's what I would say. Augmentation. Is that a good example?

Speaker A: Yeah. Great. Perfect.

Speaker B: Okay, then if we go to like the automation and I'll get into orchestration next because like automation, agentic automation, automation, they're very similar. So you'll see like a multi step process through like let's say a narrow work stream. And what I mean by that is let's take the augmentation example of like. All right, well I want in my basically a uh, tool to look at automatically my calendar based on who the means are on the calendar. It does all the account research and then preps my first appointment for me as that uh, put in the CRM right like that's an example of like an automation but it's just focused on sales. All right now in orchestration and this is what some things like that I'm starting to see really good progress on it. There's just a lot more involved with it. But where effectively like I'll take the sales call transcripts in terms of what I have. Um, actually let me define it first. So orchestration is like a combination of, of basically agentic workflows across different work streams or different departments. So with also like learning on top of it is what I would say you could do learning in agentic automation but learning for the model to get better and better as it goes on. So an orchestration example would be let's say I'm doing sales calls or my, my revenue org is doing sales calls. What it's going to do is look at all those sales calls, those transcripts and then it's also going to identify um, almost like on a, on a rolling basis the scoring of the team, how effective they are, what they're doing right, what they're doing wrong. But then for marketing, so that's more from like a sales enablement byproduct of the sawdust of just the uh, call transcripts. Then you could also take the words and language and questions that are being asked during those calls throughout the sales process, feed that over to product of what could be a uh, feature product roadmap. And then at the same time you could take that same transcript and then have a voice of the customer where it's exact language and questions. So then that is a guide and starts to create content based on all the live feedback you're getting. That's like an example of an orchestration. So it's almost like a, a running while sleeping. You know like running or working out while sleeping approach where it's just happening in the background. Um, but it's multi, you know, multi system touch. Going back to why I said at the beginning, augmentation is really critical to understand that is because in order to get quality juice out of that orchestration or the automation you have to know how to prompt really well and you have to know what look what good looks like and deliver that and basically know how to use the AI models with that. A lot of people want to jump to the automation or other areas. And then that'll see. That'll be like the example of like when you send, um, an email that starts with, I hope this finds you well with 2 million people, because you don't know how to prompt the model. Right. So, um, it's more harm than good.

Speaker A: Interesting. Okay, that's really good, helpful, uh, example of those stages. So I want to make sure I don't lose another question. So I'm going to ask you this one. We're going to go back for a sec. So would do you think that orchestration is the sort of, let's say, final stage, or is there a fourth on the horizon that we don't know about yet that we're not there?

Speaker B: Yeah, there's an, I mean, I would say there is a fourth stage. And that fourth stage is autonomous organizations. So Ilya Sutskov, uh, who was the, one of the co founders of OpenAI, also was the mind and the creator behind Tesla's Autopilot program and driving program for, for them. And he's doing some other stuff, but he's, he thinks, and he's been quoted as saying that by 2030 we'll have organizations a hundred times as powerful as Apple. It consists of these. Think of it as, um, autonomous orchestrations happening with very minor human oversight. So that's kind of what's perceived at the next level. And then you could even get further beyond that with like singularity, where. Yeah, but that gets kind of like,

Speaker A: now we're getting into, you know, Robert A. Heinlein in, in this sector there. So, um, Okay, I get that. And how off, in your view, do you think we are before we get into something like that fourth stage? Are we close or is it far away? Like where do you see these sort of autonomous.

Speaker B: I mean, that guy is really freaking smart. And I've seen other like, projections by AI researchers. I mean, I could see stuff like that happening by 2030, which is scary. However, the one part that's kind of a blind spot to me right now is the constraint. And I'm starting to see more and more about this with the models and electricity and infrastructure. So I think like, like for example, agents were available to be used way earlier than they were released. So I think like, there's an element of that as well, right? Where like, there's a lot of infrastructure. That's why the government committed to $500 billion in the US on AI infrastructure. Right. So I think there's some scaling that needs to happen like that to be able to Support it.

Speaker A: Yep. And I'm seeing a lot of crazy investment going on in these big energy sectors now for this. And it's crazy. And also I see that the energy consumption is being drilled down too. They're making these things more efficient and use less energy and they're trying to balance things out. But it still seems crazy. I read crazy article about just how much electricity like one instance of a GPT conversation takes up. It was some weird guy, like he broke it down to like how much energy you're using to have someone compute some stupid thing like do a search for a restaurant on GPT. Anyway, the point I'm making is I hear the constraint there and I'm fascinated to see where that goes. Back to my earlier question though, I'm curious to understand, like when you look at these things like augmentations and implementations around orchestration and whatnot, is it such that you think that these things are going to. Because I know this question is just so dry already, but what types of people do you think are actually going to be sort of out of work? And what people are, do you think are going to actually become enhanced by knowing these things? Like, I'm sure there's going to be benefactors of this, right? People who are smarter about these things and sort of know how to work alongside them. And then there are people who are just simply going to just not be needed anymore. Is that true in your world? Do you think it's sort of like a little bit mythological?

Speaker B: No, I think it's totally true. I think, um, and here's a, here's the interesting or a good analogy. It's kind of like how Blockbuster, you know, created their own obsolescence, right? With Netflix, where they had the chance to buy em, they had the chance to be them. They started to be them and then they're like, nah, we're losing too much money over here. So it's the same thing I see happening with individuals is like. And um, it's funny, I've seen really wealthy people have this, that have no clue about AI or what to use with it. So I think what's going to happen is you're going to see a ton of people with um, massive success already or massive wealth. They're going to get disrupted. And so you're going to see disruption there. I think I've seen discussions, I think it was Deloitte that did this study where they're predicting by 2030 less than half of the Fortune 500 are still going to be in the Fortune 500 or even exist because they failed to adapt fast enough. And so that trickles down to culture and people. And what I think is going to happen is there's a lot of low level jobs now. And I'll give you, I'll give you one release really, really quick example. Let's. We're doing a podcast right now, right? So I used to pay a podcast agency $1,500 a month or whatever to process all the podcasts, edit all of them, you know, do all that work for me, thumbnails, all that. I could literally do the same thing like 95% of the way there with a $30 piece of software, um, that's AI enabled, that has that capability. And that last 5% is human. Mhm. So like that's one example and that's like a whole industry of like podcast agencies, right, that do that. So there's a lot of different work products like that. And so people that don't adapt and like continue to level up their game where, you know, they don't operate in the old way, they operate in the new way, then those people aren't going to be necessary anymore. No one's going to want to hire. So what I see though is the biggest lie being told right now, a lot of it's by software companies, is that you're not going to get replaced by AI. Uh, AI. You're going to get replaced by someone using AI. That's true, but it's also bullshit because you will get replaced by AI. And another example, and I'll tell you this is directly related to CROs. I was sitting in with the CRO and I was walking them through how to use deep research in ChatGPT. And effectively they wanted to move up market, they wanted to go in a new vertical and they wanted the language, the ecosystem, the competitors, the pricing model of those competitors, Blue ocean areas, all those areas, right. So I set that up, Warren, and in less than 10 minutes basically had all those answers for him customized to his exact company and what they do and what they sold in 10 minutes, he's like, Ryan, I would have paid $50,000 to Gartner last year to create the same report. So if it's doing work like that, I mean, like, let's be smart about it, it's going to be able to replace people, right, if they don't adapt.

Speaker A: So I agree. I certainly do think without doubt the consulting and research organizations are going to be, you know, kind of decimated. I mean, I already see it myself, how quickly I can get things done with my remedial understanding of these things. So, so I suspect that that would be what you'd say. But how about salespeople? Do you think salespeople are going to be replaced by AI and if so, which ones?

Speaker B: Yeah, I think there's going to be an element of that. I think the bottom third, I wouldn't be surprised if, if they got eliminated. When I say bottom third, I mean in terms of like dollar transaction value, like deal size I guess is, is a way that you could put it. And the reason why I say that is um, an area that really needs to be focused on is because like these models are and tools are enabling you to do like one to one personalization at scale. And so if you have that capability, that's a lot of what a salesperson's job is to do one to one personalization. And if the dollar amount's low, there's, there's going to be comfort in using a machine to do it if it's done right and effective. And I could see that trust going up over time. Right. So I think people are still going to want their handheld for complex decisions in other areas. But I think like the area most ripe is that bottoms that are that focus on really small deals with high velocity.

Speaker A: So let's talk about prompting. You mentioned it before, which, you know, I had a really cool conversation. So I built a Delphi kind of uh, version of myself. You know, it took me a time to do it. Yeah, it was pretty cool. I uh, work with somebody on it, but I was involved in it. And you know, we built this vers, which you know, it has this whole backend all my knowledge base and all this stuff and we built it for use cases and stuff. And it's actually pretty cool. I mean, I don't know if it'll replace me, but I can actually put it up there and I can have someone ask it questions. It'll probably answer every question that they want before they'll say, all right, fine, I'll talk to you. But anyway. And it sort of sounds like me and all that stuff, but my point I'm making is the guy from Delphi saw that I did this. I'm sure he saw a bunch of people did it. And to his credit, you know, he reached out to me, the founder, and he's like, I want to talk to you about this and just see how it's going. And you know, he wanted to learn. It was really great. So I got on the phone with and we were talking about this notion that, uh, I would love your perspective on this. Was that. My perspective at the beginning of this call, it changed after the end of it. But the beginning of the call was a product that requires that I have to learn how to talk to. It is not a finished product because I should be able to just talk to it like I talk to anything else. And it should be. Be able to understand me and learn me and know how to speak to me. Much like, I would say, like Jarvis. Right? I mean, Tony Stark doesn't need to speak specifically to Jarvis. It just talks to each other and it knows it wants. Right. Because I think that that's what maybe the standard, you know, human like me wants this thing to be like, I can just talk like myself. And he had it for a really different perspective. He was like, nope, this is a new language and you're gonna need to learn a new language. He said, if you moved to another country that spoke a different language, you wouldn't wait around for them all to learn English so you could speak normally. You'd learn how they speak, and that's what you gotta do. And he said, this is a new way to talk and a new way to communicate. And people younger than you aren't having this conversation. They're just learning how to talk this language. And they're gonna go, and there's gonna be a whole slew of new people coming into the marketplace that are younger than you that aren't gonna be confused. They're gonna know how to talk to these things. And that's just the way it's gonna be. Because these tools are always gonna require a certain type of structure in the way that you, uh, direct them to do things. And I was saying, like, won't. Won't it eventually just kind of close? Won't the kind of vector go together and eventually it'll learn how to speak more human and that'll go away? And he said, probably not.

Speaker B: I don't know.

Speaker A: What do you think? I mean, is prompting going to continue to be one of the kind of last mile skills that people need to have to get these things to work, or will it just become something easier? Also, why don't we just ask the tool what the prompt is based on what I want, and then have it ask itself that question? Like, can it create the prompt on its own? I'm just curious what your thoughts are on that particular way in which we engage.

Speaker B: Yeah, it's a great question, man. And like, I guess, like, I'll look at it through this lens and it's funny because like there was uh, I don't know if you remember the movement of like okay, well prompting is going to be dead. The models are so much better. You're not going to need to know how to prompt. Good. Right.

Speaker A: I heard that. I heard that. I heard the opposite too.

Speaker B: So I remember that being really prevalent. I don't know, maybe it was like six to nine months ago. But here's what I, here's what's really interesting as these models change, what I've realized and I'll give real like tactical examples is that uh, the way like it's not about how the model always communicates with you. It's like how do you communicate with the model to get the best possible result? And so like if you look at it through that lens, it kind of reprioritizes like how you approach it. And so like, like what I've seen of my own personal evolution, I remember I had to do very basic things to get decent outputs, or I should say more involved things to get decent outputs. And it was very specific. I think a lot of that's gone away however. Like have you heard of JSON prompting at all?

Speaker A: I have, I um, have, yes.

Speaker B: Okay, so like for example JSON prompting, that's an example. It's like, it's a type of JavaScript, right? A uh, language basically in terms of how the large language models are trained and there's back and forth on like is it more useful or is it not? Well here's the thing man, like I'm, I was able to create a hundred thousand dollar videos in Google Gemini because I was JSON prompting and like if I didn't know how to JSON prompt, it would have been nowhere near as accurate or close as what I wanted the desired result from that AI model would be. So that's uh, one example. Another example is like there's ways with GPT5 and it's so funny because GPT5 is getting bagged on. I'm actually talking to an executive at uh, OpenAI later on today. I brought this up to him. But like one of the things that I said is like GPT5 is getting banged on. And the reason being is because like people don't understand the nuances how to use it. Because like you could effectively use GPT5 like an agent without all these complex nodes and workflows and connections if you just know how to prompt it the right way. Right? And actually you know what dude, I'll give it to you if you want. I could, I could Put that as a resource for people. There's, um, there's like a cheat sheet I created with prompts on it, and it's got it for CRO sales and marketing, please. Would that be helpful?

Speaker A: I'll share with my listeners. I'll put it on LinkedIn for sure or something.

Speaker B: Yeah. And you can take a look at it. It's got like the 10 different types of examples of how to prompt into like GPT5, um, with basically exact use cases for CROs.

Speaker A: That'd be worth the price of admission. So I'd love that for sure.

Speaker B: All right, well, I'll share that with you. But that's an example. That's another example. But here's what I would say to round it out. Those two areas weren't possible before, right? Like two years ago, if I tried to prompt like that, it probably wouldn't work. But now I prompt that way and it's giving me results beyond what I'd ever dreamed. So I think the bar is going to keep moving like that. Does that make sense?

Speaker A: It does. It does make sense. You know, I became completely changed after this conversation with him. He was very convincing and he made a very good point. You know, he could, uh, in a way, like, he answered me, show me how antiquated I am, you know, to think, like, somehow, you know, like I, I have to learn. I have to learn skills. And I, I get that and I'm fine with it. But, you know, the resistance I have is this sort of myth, mythological vision that we've had in fully science fiction movies that we just talk to robots like they're our brother and sister and they just do what we want, when in fact it's probably going to require some different ways of thinking and stuff. And, uh, I do agree with you that even listening to this one conversation we're having right now is if you're an engineer and you understand that stuff, you're probably gonna even have a bigger advantage because I'm not one. Right. So I couldn' Use JSON language at all. I don't know what you're talking about. Theoretically, uh, I do, but I couldn't do it. And you know, do I want to catch up or not? I don't know. I think most people probably wouldn't, you know, But I do think that this skill set is one that no matter what age you are, you need to learn it. And I would say let's more specifically talk about Chief Revenue officers. So I had a really good conversation with the Chief revenue officer of Spotify it's really Shopify, Shopify. Bobby. Bobby Morrison. Great, great conversation. And, you know, he. He says, listen to it. It's a great episode. He said, like, how you know that the future, the current, maybe modern CRO is a nerd. You're not like some kind of, you know, big revenue leader that comes in and, you know, big swinging dick kind of thing. You know, you're like, really more of a nerd. You have to know how to build systems and models and understand how these tools work and automate things. Because today's CRO really is me, someone who knows how to set this stuff up and make it go to work for them in ways that are going to make your company work a lot faster in today's marketplace. The CRO profile is changing. And, um, that is a big thing that I'm trying to get across to the people that work with me is, you know, I have good skill sets around, teaching you a lot of things related to revenue growth and revving, leadership and all that sort of thing. But when it comes to the technology and how you implement it, it's not just buying a tech stack anymore. It's not just Salesforce and a bunch of plugins. You know, today it's really one CRO that knows how to do this, can be incredibly effective today, more so than ever, in a much quicker, fast pace. And I think the ones who know these things, like you just gave that model before, how you did something in 10 minutes that no one could do in a month or whatever, those are the kind of things I think are going to make a big difference and to deploy those things across an organization. So this is the thing I want to get into with you is so I'm a CRO, um, I want to learn this stuff. I know that you, you do this. I know we're in the process of working on something like this, but the difficulty I see is, okay, I know how to do it, but how do I push this across my organization? How do I get it integrated into this as a new skill set? And the barrier there I see a lot of companies having is they can't get past that. You know, they're already having a hard time just in general sales enablement, let alone AI enablement. So what's your kind of general advice to a CRO who wants to adopt these tools and then get them working across their teams?

Speaker B: Yeah, that's a great question. And, um, it's tough, man, because I work with CROs all the time and CMOs and now CEOs as well, like a lot of the CROs look exhausted.

Speaker A: Yep.

Speaker B: Right. Because like they, they got really aggressive jobs. They have tons of mental calories. They burn on things that aren't, aren't they, that are just a part of it, that are reactionary. Right. And it's hard to get ahead of the ball in terms of strategic, if you will. So like I think the, because it's changing so fast, everybody wants that immediate, immediate result, immediate action, immediate transformation. And like, like you said, I've seen actually with executives like just a massive difference in two weeks. But if you're doing it on your own, I think like it really starts with like um, the basics with yourself. So it's like you gotta put that oxygen mask on yourself if you want your organization to do it. And just, Even if it's 10 minutes a day, just start using it or trying it for things and just talking to it like it's a human and telling it what you want. And in seeing what happens as a byproduct of that and what happens, this happens all the time is people's curiosity grows and the results grow at the same time. But it's gotta start with the leader with implementations I've done. Cause I'm, I've rolled this out to probably over 2,400 different executives now. The ones that do really well organizationally that have like, I've worked with companies that have doubled in size in a year. Mhm. Uh, and it's because the leader embraces it and also kind of uses a prompt on a daily basis of like when their team comes to them with a problem, you ask them like, all right, um, what did AI tell you about this problem? What kind of feedback did you get? Right. And they're like, well I didn't do that. It's like, okay, go back and do that. And then they're like, all right, well I still don't know what to do. It's like, all right, show me the chat thread that you had. Like how do you. So it starts to ingrain that. That's like their first level of attempt at trying to solve a problem outside of thinking for themselves. So I think that's step one. And then so once the leader has started to do that, um, in terms to help their team adopt, the other thing I would say is ask for a couple hand raisers on your team, have them tackle one to two identified use cases, have them create prompts or GPTs or custom assist and then present the results back to you in two weeks to present the results. Then you have the team test it out or a portion of the team, and then basically you have them give their feedback again after two weeks. So, like, that's a very, very simple way to start and then you gotta compound on top of it. I mean, I help companies roll this out, so there's a lot of this I do for them. But that's what I would say is a very easy way to start. Just to get some kind of a mic.

Speaker A: I got it. So you sort of have to just. Just get the ball rolling and keep interacting with these things to the point where you start to feel comfortable with them. I mean, it's very similar to what I've done. And I feel much better now. I'm nowhere near where I need to be, but I'm getting closer all the time.

Speaker B: That's awesome, man.

Speaker A: Yeah, it's getting there, you know. Believe me. I'm just curious. I like this stuff. I'm a nerd, but, uh, it's a good. You'd point. And I was just saying this to a master's, uh, council group of mine yesterday, which is, you gotta have your cans on the keyboard yourself. You can't just hire some rev Ops person and say, be my m. AI person. You know, you have to be the AI person. You have to know how to use this stuff and use it in a way that makes a difference so that it brings you a different level of personal knowledge and credibility when you're bringing it across your organization. And you need to know how to do this. And it's much like spreadsheets. You know, I can have some, uh, rev Ops person put together a complex model for me, financial model. But if I didn't build the spreadsheet myself, I have a different relationship to it. And I think there's a similarity there. So, um, that's really interesting. You know, I'd love to hear just a little bit more about you. Like, so how did you get into this? You know, were you an engineer? You know, you obviously have a good brain for this sort of thing. And it's a good story to know how you found yourself in the AI space and a little bit more about your career journey.

Speaker B: Yeah, man. So, um, I was actually in go to market for close to 25 years myself. And a lot of that was as, you know, individual contributor from sales rep. And then I did that. I started off, uh, at like a boiler room type environment, like inside sales heart, you know, one call close to investment banks and brokerage CIOs. So started off doing that, uh, and Then did mid market and then enterprise, and then moved up the chain, got into leadership, did kind of the same path, mid market, enterprise, worked for some, you know, companies, had exits, te backed. I also helped a, uh, small company grow and sell a little publicly traded company that was more in managed services. So, like, I think part of the training ground for that is like, I always work for companies that were resource constrained, didn't have a lot of infrastructure in place, and always were just like, go make shit happen. You gotta have big results. Go make it happen. We don't care. Just figure it out. Right? So that just is context. I grew a group from 0 to 30 million in five years with four salespeople. Um, and then I left and started consulting companies on that. But what happened was I have a podcast too, which you're going to be on, obviously, right? Called the Scale Up Show. And I had a guy named Chris Savage from Wisteon, and he's just like, hey, Ryan, have you heard of Dolly? And this was in, I think it was like around September of 22. So I'm like, no, what's Dolly? And it was like OpenAI's first generator model, right? Which was terrible at the time. Terrible, right. Version one. But it's funny because OpenAI thought that was going to be their flagship product, not ChatGPT. Well, then ChatGPT came out. I got instant access to it. And the thing that hit me between the eyes like a sledgehammer was that I asked it something that took me 10 years to learn, and it got 95% of the way there in three questions. And that was with GPT 3.5. So once that happened, I was like, shit, this is gonna change everything. And I remember my wife and I were on vacation and I'm like, like, I think this is the right way to go for my business. And so we had to talk about it and she encouraged me to do it, and then I just kind of went all in from that point, effectively. I've been doing AI implementations for three years now, specific to go to market and executives. And so that's kind of like how I got to this point, man. And I, uh, love it. I love to nerd out on it, right? It's tiring because it happens so fast sometimes, but it's also energizing at the same time. So it's kind of a yin and a yang effect, if you will. But overall, that's, that's kind of how I got to this point.

Speaker A: That's beautiful. I love it. I got it. So you, you Saw you saw an opportunity, you recognized a pivot and you jumped on it. And I can relate to that. You know, the same thing happened to me, my career, a long time ago and changed my career dramatically. So I think it's sort of like you have to see things and jump on them, you know, And I don't know if I've been always good at that. You know, think about how many stock picks I probably missed out on. I look back on, I was like, man, I should have bought that stuff. But, um, you know, it's cool. And I think that to your credit, you know, you have the right brain for this. And you're also sitting at a seat in the beginning of something that's going to be revolutionary. And I think it's amazing. And I think like, people like yourself, whom, yeah, people like yourself understand how to do this stuff really well, are really valuable today because it's so incredibly curious and important to people and it's so much part of people's consciousness now that, you know, I think people stare at the screen looking at their open chat GPTs or whatever they're doing and they're wondering if they're doing everything correctly and they're clamoring for, for insight. So good for you, man. You know, I got a question about something. You mentioned this before about, you know, you do stuff and it, it's 95% right. How do you know? I mean, I've seen a lot of stuff being produced out of these platforms that's just wrong. Like it's, it's, it's, it's completely just. It sounds good, it's well written, it's well constructed, but it's just not correct. But then there are things that I don't know if it's right or not, and, and I sort of trust it. And so, you know, how does this all hallucination stuff get resolved? You know, when you're building things that are involving complex research and other things like that, that maybe it's just not correct. How do you know?

Speaker B: It's a great question. So I think, like, think for me, if it's a big decision or something I'm relying heavily on the model for that has a really big impact. Then I'll double check it, you know, uh, multiple areas, check multiple sources just to make sure that it's super dialed in. Supposedly with GPT5, hallucination rates dropped down to like 4.7%.

Speaker A: Okay.

Speaker B: Which is, you know, pretty freaking good because like, I don't know about you, but like, if you look at it from a human perspective. I, I would say the human hallucination rate is probably like way, you know, maybe 20%. So like, so that's like, that's one thing you gotta remember too. Like most people get wrapped around the axle like, well, it's not perfect every time. It's like, well, neither are humans and we rely on them all the time and they're probably substantially lower. So like that's something to factor into. Um, and then like if it's outside of my domain of expertise or things that I have an experience with, then I'll also do further inspection to validate that. That that's what good looks like. Right. There's also other ways to validate what good looks like too. Like, and this is a great example, but in social media content, right, you could look up posts that really do well. And let's say you look at one for a small creator who had a post that did really well. You could feed it in there and have a reverse engineer. Like why did this do well from a copy framework, spacing, messaging perspective? And then you could start to reverse engineer success too. So that's another way to do it.

Speaker A: Yeah, yeah, that's true. And I do, I find People's posts on LinkedIn that I liked and they did well and I pump them in and I say, right one for me, that reflects the same principles at this one, you know, so I do that a lot. Um, and it's good, it's effective, it works. Very interesting. So, uh, how do people get a hold of you? Who are you looking to talk to? What type of things are you doing for people? How can we get more access to you and your, your unique skill set?

Speaker B: Yeah, man. Well, I know you and I are cooking up a partnership.

Speaker A: I know that. I'm excited about it and it's going to be very cool and uh, I'll be telling everybody about that soon. But yeah, Ryan and I are working on something together that's going to be

Speaker B: very cool, something special. So that's one example. The uh, other thing too is, I mean, feel free to reach out to me and connect on LinkedIn. I publish content daily on AI specifically for go to market and I'll give like tactical use cases, examples. I'll provide that resource link so you get free.

Speaker A: Yeah, that'd be great.

Speaker B: That are CRO specific and that's usually who I work with. I usually work with a CRO or sales leader of an organization and basically help identify the KPIs or the constraints that they want to solve with AI and then point specific use cases, solutions to transform their organization in 60 days. So that's like my initial flagship program that's done really well. And I've never had anything that I've ever sold in my life or worked with where I've had happier customers with more inbound referrals. So lets me know that I'm going down the right path. And I love seeing people get the results, because, you know, one of the hardest things for me is when I had a job and it didn't work out for me. So if I could help people be proactive about their future, their org, and what's possible for them and their families, and that lights me up so that they're not in that position that I was in when I got, you know, fired unexpectedly. Right. For a position.

Speaker A: So, amen, man. That's exactly why I'm doing this, because I like helping people and it's much more rewarding. So good for you. And I think it's great. I think what you're doing is amazing and, uh, really impressive. So thank you and thanks for being here. It, uh, was a great conversation, and, uh, we'll be talking again. So, uh, Ryan Staley, uh, founder, whaleboss. And, uh, you'll be hearing from the both of us soon. So what I'm gonna do now is I'm gonna do the intro and they tack it on the beginning. And I'm not using a bot to do this. I probably wanna find out what you're talking about because I am paying somebody to do this for me. So. Welcome to this episode of the Sierra Spotlight Podcast. This is Warren Zen. I'm the founder of the Sierra Collective. And, uh, you know, the ongoing conversation we have every two seconds is AI and how it's being used and how what works, what doesn't work, it's. It's the obsession of our decade. And I hear about this a lot because my. My partners and my clients are all talking about how do we implement this stuff, what works, what doesn't work, the amount of confusion, hype, and, uh, you know, I had the recent, um, pleasure of being connected to a gentleman by the name of Ryan Staley, who is amazing, ah, practitioner at AI company called Whaleboss. And. And I heard about him through a client of mine, actually, and then, uh, asked Brian to be a guest on my CRO accelerator course. And I was just blown away, as I suspected I was. In terms of Ryan's knowledge and application of these tools and the sophistication of them and how. What's possible. So I asked Ryan to be on the show, and, in fact, he and I are going to collaborate on something, and we'll get into that at some point in the future. But, uh, I'm really excited to welcome Ryan Staley. Warren, thank you for being here and looking forward to talking to you.

Speaker B: Yeah, I'm super excited. Warren. This would be a lot of fun, man. I love what you're doing with the, uh, CRO Collective and who you serve, because it aligns with who I serve. So really excited about this episode.

Speaker A: Great. Thank you. All right, so let me just stop it.