AI is everywhere in business conversations, but practical implementation still feels unclear for many CEOs, leadership teams, and B2B companies. This week, Bill talks with Josh Hochstedler and Aaron Perry, co-founders of Ruper Labs and the team behind 50 Marketing’s new associated brand, AI Implementation Specialists, to discuss what it takes to make AI useful inside an organization.

Our conversation moves beyond AI hype and focuses on realistic implementation: small pilots, human-led systems, process mapping, data readiness, token costs, modular AI architecture, and measurable ROI. Josh and Aaron explain why most companies should not begin with enterprise-wide transformation, why AI needs the right human operators in the loop, and how businesses can identify the right seats, workflows, and use cases to create meaningful gains without overwhelming the organization.

Bill: Thank you for joining The Missing Half Podcast, where we're discovering what's missing today in AI. I'm joined by some very special guests, some gentlemen from Ruper Labs. Aaron, Josh, thanks for joining us.

Josh: Thanks for having us on.

Bill: So everybody's talking about AI. No one understands its full capacity, its full capabilities, but you're trying to bring some common sense approaches to actually winning with AI, with our new venture, AIS, AI Implementation Solution. So let's talk about, what we're going to do there and maybe unpack and hopefully learn together and learn with our audience what we can realistically expect to do with AI.

Aaron: Sounds good.

Bill: Great. So you guys have been participating in this space for a couple of years. Which is a lifetime in AI, right? We've been doing this for about 12 months, 14, 15 months, and we're seeing, I think really a better understanding of what happens with AI. Maybe talk about the difference. Josh, yesterday we talked about transformation as opposed to small steps to win.

Josh: Yeah. No that's a great topic. And I think that's something that we're seeing a lot even just in the news is we're seeing all sorts of big companies that are saying we're going to transform a company completely. We're going to lay everybody off and AI is going to do everything, and it's going to be great. And inevitably, we've already been seeing people starting to hire back, people starting to look for, incoming people to join new positions and fill those roles. And I think that where those things go wrong is that change, especially at an organizational level, and once you're at a bigger organizational level, needs to happen incrementally. And so this AI is going to transform everything, AI everything, isn't necessarily the best approach.

Bill: So Aaron, another thing we talked about yesterday, in our hackathon session was kind of this concept of 80%. We're not seeing AI able to really develop solutions that are 100% agentic. But if we have human-led 80% solutions, we can really show some tremendous gains. Maybe talk about that a little bit.

Aaron: Yeah. For sure. I mean, there's, Josh kind of voiced there. There's a lot of over-promise in AI. And I mean, even touching back to what the question I asked Josh and Anthropic is introducing something corpse. And it's like where they introduce these quote unquote AI engineers into different businesses paying them 85 grand a year just to go in and make these small steps. And so they're seeing even Anthropic is seeing that the importance of having a human in the loop. And you said, kind of something we were talking about yesterday, you said, as much. And I pointed out that that is a very, I think, crucial phrase within that sentence. As much people don't need to be in the loop, right people still need to be in the loop, but not as much. So there's still these processes that AI could fast track or that 80% the repetitive tasks, but that 20%. You still need those individuals who are competent who, trust AI, but they verify, right? Trust but verify. And so having that individual in the loop to be able to see the AI’s output and then check it and make sure it's verified before sending it to the client or higher up, right. That's a super crucial component of this whole AI infrastructure that we're trying to implement within business.

Bill: So when we hear about what's happening, any every news clip, every, business news wire, everything on social media, it's transformation. Change the world. Dystopia, guaranteed income, all those things. And whether that's true or not, I'm not smart enough to know we're not smart enough to know in the future. But when we think about the practical steps we can recommend to clients every day, when we think about the things that we can actually impact this week with our clients, maybe we should take a little bit of time here and talk the audience through how we're bringing really practical improvements on an incremental level, at almost a seed level to, companies’ operating systems or their development or their growth. Because that, to me feels tangible and less, less of an academic conversation or, a futuristic, hand-wavy. Yeah, yeah, yeah. Invest in me, drive my IPO. So maybe, let's talk yesterday. So, just for, the audience’s sake, we had an all day hackathon yesterday with Josh and Aaron and a couple of our other team members because we're really trying to build a company that is going to solve these problems. Not only for ourselves, but for others. We’re having great conversations with some of our current partners. And we're really seeing in incremental improvements are where we need to focus to win, not overall transformation, and let's talk about some of those problems with overall transformation. And we outlined a couple of them yesterday. Josh, you talked about organizational friction and some projects we've started with companies that have lofty goals. And we're six months in and nothing's happening because all the friction internally with their corporate, roles and rights. So maybe talk about that alphabet.

Josh: Well, I think that what we came to through our conversation yesterday was actually the importance of the relational side of things, of organizational change begins with people. And so if you don't pick the right people and start with the right people, then everything just kind of drags out, right? And so I think that going forward, that's one of the most important realizations for companies, is that they need to identify who within their different departments, are ready to drive those changes and ready to be interested into stepping into what AI can offer in terms of improving efficiency and cutting out boring manual tasks without being afraid of job replacement and stuff like that. And so I think that the, you begin with identifying your entry points into your processes, and that is the people that you have. And then I think there's a very interesting parallel here with software engineering and what we're seeing in AI in that space. So as I've been researching that and kind of working through those things, it's been coming to light that people are going back to what they were talking about ten, 15 years ago where there were these guys writing code, writing books on best practices for code. And so it's like, you know, this is how you should approach these problems. This is how you should engineer an architect, these things and these AI guys are saying, we need to focus on that now more than ever, because that gives the AI the structure within which to operate. So taking that idea then and applying it to different department departments, I think it looks like figuring out the best practices for different departments, implementing systems, which we all already know that we need to do at some points in our companies.

Bill: Right.

Josh: There's already some sort of, idea that there's always optimizations to be made. And so taking that and applying it through AI and speeding it up through that way, I think that's where you get those small incremental but exponentially returning results.

Bill: No that's great. And I think we also talked about data being a problem. Right and where we're struggling and where we've seen projects just absolutely hit gridlock is where we we working with the company. And then the data organization and creating a data lake becomes a gridlock. Where we're finding success. And we talked about this yesterday. Aaron maybe you could add to this is where we just pick a seat, get the data for that seat, deal with that data instead of the enterprise-wide data optimization or transformation. We just need to know what that person does, how how that seat functions. The day-to-day impacts of that seat as opposed to creating this monster database. Maybe talk about how like if you go after transformational change, it takes forever and it slows down, where if we can just deal with that seat, we can really move things forward.

Aaron: Yes, definitely. Yeah the thing that we I think we realized with seats is this adoption curve. Right. When you're identifying a seat, there's always, you know, we have early adopters, late adopters, anybody in between. And so you’re looking for a seat within a corporation to have that kind of of the transformation. But that small incremental progress with AI, you have to find those individuals who understand the importance of AI, and who are those early adopters, because then if they're excited about it, they'll text you. Or if we're working with them, we'll probably text to say, hey, what if we did this right? We want those type of people to work with because they’re fun to work with. And we can kind of had that that collaboration. But considering the data component. The nice thing with having structured data with individual departments is data looks differently per department. So if you have sales right, you have a lot of sales transcripts. Well that’s just text. AI loves text, right? And if you're in logistics and it'll be a lot of numbers, some words as well, don’t get me wrong. Maybe in marketing there's a lot of images and AI is getting to a point where it can really understand images, but it's not fully there yet. And so say we could just focus on sales, train a specific, agent or an LLM or whatever it might be on that data. Then we have something that's specialized with a specific sales department that's not trying to pull all this other data from all these other departments with different types of data. Right. What does that do? Well it makes your token usage a lot more efficient and cheaper because you're not using as much tokens. You're pulling from information that's the same information pretty much every time. It’s the same type of information, same type of data. It's not using as much tokens because AI knows exactly what it's searching for. It's not searching for all these other different departments with all these other, other different data sets, we're able to be a lot more efficient because you're focused on one specific problem within the company and not trying to transform everything at once.

Bill: Well and I'm glad you brought that point, because we we discuss and describe data as a monolith in every organization, right? Oh, it's just one thing. Data. It's just data. Yeah, yeah, just go grab that out of the filing cabinet from my generation. Put it in the the computer in the server or whatever. We're good to go. But I think you brought up a good point. When you're starting with transformational change, and you have that as your thesis and you're looking at data, all data is not the same. The ability for AI to interact with different types of data. There are different maturity curves that exist there. Right. Because video and photos, images, that is way behind where it is on ones and zeros and text, right. Or MD files. So if we, we're going to be forced to take small steps because you can't just take and we've seen projects like this. Right. Take all the data, put it into the thing, right, Google Drive, some type of server or whatever, and then just attack it with an LLM. And we have horror stories of some clients who come to us after they have done this and racked up huge bills because their team decides to do whatever and make these erroneous requests to just churn it, churn it, use the tokens. So I think when we look at that, that data issue, if we're if we have the humility to accept where we are in this moment with AI, that it is an amazing opportunity, an amazing tool, and we can get ten x returns. Like, like, not no other time in history, which is what makes it so exciting. But we can't do that across the entire organization simultaneously. We have to pick our spots. We have to, pick a department, pick a seat, pick a small data set that we can make sure that we can train the agents, the LLM, whatever on. That's where we can show the greatest wins. So I think that's that's a great point.

Aaron: But just to add to that, and I actually shared this story yesterday. So this is Daniel Priestley. He was on The Diary of a CEO with Steven Bartlett. And he shared how within his company he had a big sales department, and they decided it to hone in on AI within their sales department. And they analyzed all their transcripts, all the call transcripts. First time they had done that in a very long time. And what they identified was a lot of the objections that they were getting were having to go talk to somebody inside the corporation that was also a decision maker. So you have one decision maker on the call, but not all of them that were necessary to make that decision. And so whether that's a wife or a co-founder, whatever it was, and he realized that within that whole kind of process, getting him to that call, they didn't ask that individual if they wanted to invite anybody else on the call that was a decision maker. And so once they did that, they said, hey, invite your wife, invite your co-founder, whoever it may be onto this call and let's have them both on let's let's kind of bang this out with one call. And once they did that I think their conversion rates hopped up by 30%. Right. And that's a huge jump. And that's directly tied to revenue, right. 10x because a lot of people are saying 100 X like we're being realistic. Ten x, if you have 30% higher conversion rate on your sales calls, you will have a decent amount more revenue on top of that. And so looking at a specific department, in this case sales and tying that to, honing AI into that, and having a lot more efficiency within the department. You'll see a lot more returns instead of having like you said this model of data trying to analyze everything at once.

Bill: Well and isn’t this the lie? That and maybe lie is a harsh word. When you look at the AI companies, they can 100 x, they can a thousand x. SaaS companies can do that. And so they talk about those type of gains because they can achieve that in that digital world. When we think about physical businesses, when you think about manufacturers, wholesale companies, distribution companies, industrial service companies, if I went to one of the CEOs today and say, hey, we going to 100 x your extra business, they will go pale. They will like start to sweat because one, they don't want it. Two, they know it will blow everything up that they built over the years. It would be absolute chaos. And most of them are doing well enough and they just want to keep making incremental changes and be on the cutting edge. They don't want to be on the bleeding edge. And I think if we can, bring that type of common sense to the market, that'll really separate this, irrational exuberance that's on, all the business talk shows driving all the valuations, you know, so that’s Wall Street. We need to bring this down to Main Street. We're just plugging away day after day, making wins. I mean, any any CEO in America, if you can tell them we could do a project that could increase sales conversions by 30%. They don't want to hear about transformational organizational change. They're like, I want to buy that one. Right? Can I have that à la carte, please. I don't need the rest of the menu. We're eating light today at lunch. Just the small, small entree, but no. So that I think is a great point. Josh, one of the other things we heard probably 60 days ago was this fear that tokens were going to all of a sudden be kind of expensive. Right? And, we're there, right? We're starting to see more fees on tokens. So we're already seeing companies. I was with a company earlier this week at lunch, and they had already shut down a major AI project because their token burn on, exuberant employees who were willing to ask, the internal AI anything was causing, token churn. Do you feel like we're at that moment where companies are going to start pulling back on just the irrational attack in AI, and they're going to require someone to come in and put together a plan with guardrails that makes sense that’s actually achieving outcomes?

Josh: Yeah, yeah, there's a lot of different angles you could take there, because I think that, to to bring in a real world story. We've already seen this at some of the big tech companies, right. Running internal leaderboards, I told you about this, running internal leaderboards for who can use the most tokens, which essentially tokens are money at a certain point. Right. And so we're running internal leaderboards to see who can burn the most cash. And there were these engineers that were developing systems that weren't doing anything to use tokens because you have to be on the leaderboard or you get sacked. Yeah. You know, and so there's a there's an important side lesson there about like how you measure outcomes, how you measure results. But bringing it back to what you were talking about, about the what does it look like to be aware of the economics around that? I think that we will start to see the app become more expensive, which just means that the slow and incremental and the strategic becomes even more important. It's not going to be about who can implement AI in most things, it's going to be who can implement AI the most efficiently. And that's where the key lies. And to be able to do that, you have to slow down. You have to zoom in and say, okay, we can isolate this use case, we can isolate this person and their process. We can say these are the inputs and the outputs, and we can deterministically to sort of extent, map that and just say step by step, okay, that's what it is and that's how much it's going to cost. And we're not going to wake up to a $500,000 API bill on Monday morning, you know. So.

Aaron: Well on top of that, you talked about being platform agnostic. Right. So say say Claude skyrockets in their token usage. And our clients are using Claude. But Gemini's got something down the road that's cheaper. Hasn't skyrocketed. Right. We can implement Gemini into their their system instead. Or say all of the different platforms skyrocket their prices because they're no longer being backed by PE, right, we can implement a custom LLM. Like it's going to cost more, don’t get me wrong, I'm not saying that there's going to be this super easy transition to that, but right. Having a client that can be platform agnostic allows us to move really fast, right? And we can we can pick and choose what's the best on the market. We all know, the businesses know. We know that AI is moving fast. You don't have to be in AI to know that it's a hockey stick. Yeah, right. And so being able to, like you said, take a step back, be simple in, applying AI, but then also being agnostic and applying AI, allowing yourself to choose all these different platforms to be able to plug in and say that this is a lot more efficient. This might cost a little more, but it's certainly better. Whatever it may be allowing the client to choose so they go with the fast, expensive or cheap and slow whatever it is. Right? Allowing us to be agnostic in our limitation is something that's will be really important.

Josh: Yeah. And the intelligence is going to become a commodity with time. And so I think that actually even we'll see what happens with OpenAI and Anthropic. They’re on a ticking time bomb to a certain extent where all they have is the models and they don't have the hardware, they don't have the other stuff. And pretty soon everybody's moving towards a model of building processes that are plug and play where the AI, the intelligence is basically just an engine. So in the same way that you can swap an engine out in a car, you should be able to swap out your AI and your processes will need some tuning. Will need some adjustments, but it should still work.

Bill: Well and if the hypothesis was, that was perpetuated for a very short period of time, and it’s kind of funny, because I think AI actually found the truth and busted some of the hype around the fundraising cycles. The hypothesis was spend as much, use as much AI. It will force. It will just like suck you into this black hole of transformation. They've already disproven that because they've given like the leaderboards. They have incentivized behavior where it was spend as much as you can on tokens, put AI everywhere, like just infuse like. You know, drink it, breathe it. Like put it in the back dock of the building that the the employee breakroom, AI everywhere, right. That didn't work. Right. And actually it was counterproductive because they spent a lot of money. And then they found out, I guess, with some of these leaderboards, that at the end of the days, the engineers would type in, run all the recipes that Martha Stewart ever made, and it would just sit there and churn overnight so that they would use tokens so they would move up the leaderboards they went. So a lot of, people behaving badly. But, so, anyway, I mean, I think that's just interesting the way that's kind of all playing out. And the beauty of it is we're getting the truth faster. So AI is actually accomplishing its goal in that where the intelligence is, maybe a counterpoint to this irrational exuberance and this hype play. But I, I want to go back to a word you guys use. And I think it's so very important that we recognize this as we take a humble approach to small steps of implementation and growth as modular. I think that is going to be one of, it's not being talked about a lot right now. But the strategy of having modular components and whether that's on your data side, whether that's on the models, whether that's on your agentic systems or your orchestration layers or, the governance, like whatever the layers are, they have to be modular because they are not, the development is not all moving in steps. It's like this one goes up, this one goes down. There's new, like right. So how do we see that as being one, adding complexity. Because you need to know the old system. You need to know the new system. You need to know the transition steps. But also being kind of a de-risking investment decisions for CEOs because they know it's not all OpenAI. And if OpenAI is a problem, we're in trouble or it's all Grok or all Claude. And then all of a sudden, poof, we got to start all over again. Maybe talk about how you guys see that coming together and why modular is so important.

Aaron: I think, I don't know if AI is at the point to where one AI in and of itself can be at an enterprise level, and so I don't think OpenAI is there. I don't think any of the platforms are there where you can just implement X, Y, and Z LLM. And you're good for that, that multi million dollar company, whatever it is, like we're not there yet. And so this modularity component to what we're trying to bring to the market allows it to be enterprise, because we can pick and choose what's operating best. Right. If if OpenAI goes down, their servers go down. in this company. It needs whenever we both for them and most likely is the case, right. We need to be able to plug and play Claude to do it, or Gemini or whatever it is, and it allows it to turn something that isn't enterprise yet into something that we can actually, implement within enterprise company because it's just it's just not there yet. And like, even though it's a hockey stick, even though we see that hockey stick, it’s just not there. And so we need to be able to bring something in the market that can apply AI into the enterprise level of the company. Even though that that kind of system isn't necessarily a market yet.

Bill: So the other thing I think we need to talk about is we talked about this 80% rule is that we want to generate systems that move us 80% of the way towards the ultimate goal, which would be, the ultimate goal would be hundred percent automation of everything, whether that's on your factory floor and you have robotic arms, you want 100% automation, but every factory I've ever been in, even with those automation tools, you still see people, right. So they haven't made it. When we think about 80% solutions, we are seeing case proof that we can develop 80% solutions with humans in the loop that, and then we can start stacking those. And that's where we're finding the wins. It's not that one silver bullet that just fixes sales, fixes admin, fixes whatever. Josh maybe talk about how you're seeing that that compounding interest idea of investment in those items really start to stack up.

Josh: It's a great question. So you start with isolating that one thing like we talked about. And let's say you implement that. Right. And you gave the case of your your assistant where she saved four hours a day. Yes. Which is definitely a bit of a unicorn. Yeah. You know, and we talked about some of the reasons why that was. What was interesting is some of the reasons was that she's super organized and that she like was a perfect case for that. Right. So that's not necessarily what you're always going to see. But even let's say it was two hours a day, you do it with her and then you have someone else in your company who's forward looking and who wants to do that and then, you saved another two hours. And after time, over time you're saving a lot of time and you're freeing up your people for more high level creative work. Right? And as you do that, let's say that the breadth of a department is like here and you have like people across it. And then between people you have your different subprocesses. So you take one process, you did it, take another process, you did it. Then you took this process, you did it. And those processes are actually adjacent now. So yeah, that's what you can you can do what we call in, software engineering or what I've seen people doing is the deepening of the modules where you connect those modules and you say, hey, these are kind of running in parallel, but they're related and they're not necessarily isolated. So we could bring those together and have them work together to reduce the complexity, because once you hit like three isolated running processes kind of thing, then you're going to run into a situation where there's the maintenance overhead, there's some things there. So if you can connect them and knock it all out in one go, then you cut down on some of that. So I think that that's kind of the idea there. Not to get too abstract.

Bill: So and Josh does a very nice job of simplifying these things down because he has an amazing technical background and experience, and he makes it so even I can understand it, which is good. Because I am not, a software engineer. I've met them before. I've talked to them, but I didn't ever open those books, so this is really good. Well, I think when we look at those 80% improvements, one, it's something that we can achieve really quickly, cost effectively and we see tremendous return. But also those are the type of returns that are able to be accepted by the organizations because we're not replacing people. And if we're working with a company that has let’s say, 20 people in that seat, we find that one person, who’s forward looking, willing to work with us, we develop a solution. And if all of a sudden we can save them one, two, three hours a day and then they can shift that time to higher value work, more value add, better customer service, whatever their performance metric is. And then we've already shown that it works. Then getting those other 20 people to adopt it is a lot easier because we're not asking them to believe, we can prove it. And then management can, you know, work to have some type of change initiative, help train other people. But we've proven it once quickly. And the reason why this is an exciting moment just in business is historically, many organizations have been unable to do that type of thing because it was so costly and took so long to find that one small use case, and then actually build that solution and then roll it out. That occurred in Fortune 500 companies and very large companies with huge R&D budgets. But when we look at the middle market, it's you had to place bets. Yeah, I think that'll work. We're going to go all in because we only have time, resource and budget to try one thing. We can go in and try something like this in 30 days and be like, oh, it's not going to work. Let’s pivot. We found this other thing that will work in say two hours a day or an hour a day or whatever the the metric is, and then you find that thing and scale it up. So I think there's this moment we're approaching that is so exciting. Which is why I'm trying to learn more of these things, that we can really make these huge changes and reduce the friction, reduce the risk, and, really get a lot of buy in, in the market.

Aaron: I think that there's this really cool benefit that you get from attacking seats. You now have someone in the business that is willing to fight for you. So. Right. We've been talking about right that trickle down effect. We hopefully see this AI being slowly infiltrated into the different parts of the business. Right. And you have that individual within a department that seat who is gung ho about AI, who then someone if you're training some people below and they can come to that individual that we've trained up on this process like, hey, it's not doing this thing right, what's the issue? We've trained that individual within the business to help train those people below them. So they're not necessarily coming back to us to ask those questions, right. Because we have those people within the company. And so there's this kind of need for and we talked about this yesterday where IT is not going to be your go to person for AI, right? Maybe for data and maybe for some regulatory things of how you get different parts of the business kind of cohesively into one spot for the data. But that's not going to be the person or the department that's going to be implementing AI. It's going to be people like us who are going to come into a business, implement AI, and then train those different seats that allow us to train those other individuals beneath them to really have that trickle down effect of AI. And that's that's some that's really cool about this kind of process of really honing in on that.

Bill: So I'm glad you brought that up. I have an article I wrote, it's somewhere in our ecosphere about that IT is not the department for this. Because when we think about what we're trying to do with these small pilots that we could scale and have really large returns, whatever the x is, we could argue over the x, but that math won't lie at the end of them. IT is a traditional cost center where they, historically, most of the folks in that space over the past 30 years have been trained to evaluate new platforms over a 6 to 12 month period. Get demos, find fit, pick one, and then the real work begins and it's ugly and everybody hates it. Takes 12 to 24 months to implement the system across the enterprise. Then we have all kinds of problems. I've never had an IT project that wasn't over budget, that was on time and that the adoption rate was high and the people liked it. Then you get through that 24 month period and then it's like, okay, now we actually have good data. Possibly. But let's just let's just let's, let's run with that.

Josh: There’s no such thing as perfect data.

Bill: Yeah. But yeah, let's let's run with it that it was good at 24 months. Now we can start to really benchmark, look at data, start to optimize and look at how do we recoup some of this investment. So when you look at an IT investment and the mindset in those departments that's been perpetuated for the past 30 years, you're looking at ten year ROI calculations. And these are these are legacy investments. And that's why people don't change their IP every other day. Right. And that's why they have ten year old systems, 20 year old systems. When we think about what AIS is going to do and is doing today, we're talking about 30, 60, 90 days, right? Like, okay, we're going to come in and evaluate, run an audit in 30 days. In 60 to 90 days, in that window, we're going to try and have some type of pilot. We're going to be testing. We’re gonna be getting data. We're going to see how it's working. And then implementation could be another 30, 60, 90, depending on the scope and scale, or longer. And the payback could start coming in one calendar year. That's that's the new math. The new math isn’t the SaaS, AI platform math, right where they just keep buying data centers and spin it out. It's can you start to see cash on cash return on an AI implementation investment in under 12 months? And the answer is yes. And we're seeing it. That's what gets me so excited because you're not selling faith, you're selling proof, you're selling hard results that you can show immediately and how that can transform. And the the size of the company, if you can find those pilots and then scale them, you can really achieve the results. And that's why it's going to require an entrepreneurial mindset to implement these, not a traditional accounting, finance, HR, IT mindset. And we need those folks. We need them to be part of the team. We need their data. We need to work with them. But it'll be a very unique organization that has folks that can sit in those traditional departments and seats who can also own the AI initiatives and not only own them, but execute them. Right. So may maybe the accounting and the IT department are going to be your liaison with us to get it done, and that's fine. But they're not going to be the ones who think AI first, think this has to make money now. Not ten years from now, not some long return curve. Like how are we implementing and then not only do we get return, how are we using … how are we, how are we reintegrating new data to make sure that it gets better and better over time? That's also one of the things that surprised me. I thought, okay, we're going to see these, okay, four hour savings, but then we're starting to see another half hour and that like another small task that we can put on or another small thing and we just keep chipping away. That's where it also gets exciting because it doesn't stop. It just keeps going.

Josh: Yeah, yeah. Done right, you can see those. You gain more insight into your company and your operations. Which is something we always knew. We always knew that more data equals you can have more insight at your company operations. But what we haven't had up until this point is the capability to analyze that data. And that's what AI is really good at. You know, no one wants to sit there and look at a spreadsheet with 100,000 rows and, you know, be tracking like different peaks and different, medians and distributions and stuff like that. But AI is more than happy to do that all day.

Bill: So you must have been sitting on my shoulders and listening to conversations because as you guys well know, we have a project where we're analyzing data set. I mean, it's not large from the greater scheme of things, but, I don't want to comb through 100,000 lines and that data set, you know, we're strapping some AI on that. We're going to do some analysis. And, I think there's going to be an amazing product out of that. One of the other things I want to make sure we're very clear with our listeners today is that when we're talking about having team members at our clients’ companies engage AI, we're not going in and helping them download Claude Pro or like, the different, $20 month subscriptions and teaching them how to prompt better, right? I mean, these are agentic systems with agentic orchestrators, and they're a quasi. If you're not familiar with this, it's almost like you're getting a custom SaaS AI thing. And I'm butchering this, but I'm trying to, you guys can clean this up, that comes into your business and sits in a seat. It’s almost like you're hiring an AI seat filler, worker, department with the orchestration to do the things almost like a machine, as opposed to, we're not just, coming in and like ten years ago, downloading and showing them how to use Chrome as their internet browser and teaching them to not just, say, fix sales. That that's, that was six, 12 months ago, right? We're now implementing agentic systems that they can operate with interfaces, that type of thing. Can you guys clean up what I just said and make it some better and more accurate?

Aaron: Yeah. I think the, the main thing that we're focusing on is, like you said, we're not we're not going to implement a software. Right. That's what the, you said people used to just come in integrate the software 12 months or 48 months or 24 months down the line, they're going to see potential ROI. We're coming in. We're talking to the seat and the individual in that seat that we've kind of identified, and they're going to voice to us those problems, right? So a lot of times, until the companies come in, they do a whole audit and we'll do a similar audit. But there's a way to fast track this with the seats. And this is kind of something that we were analyzing yesterday where the seat is so implemented within that, the department that they know the problems that the individuals beneath them are dealing with what they're dealing with. They can come to us and they can voice those problems to us, and we can know exactly yeah that that one has still got to stay with you guys like that is there's something that a human needs to stay in the loop for. AI is really not there yet. Oh, that one great. We could take you 80% there. Right. And so being able to analyze that and find it really fast is something that a seat allows you to do. And so how the software plays into that is the agentic portion that you kind of brought up is you're able to string together a lot of these different parts within the department to allow you to, kind of aggregate it into one interface. And so he's the technical guy. So I'm going to let him get even more technical with this, but a high level overview of your aggregating a lot of different components of this department identified by that seat, and implementing into one interface to allow the rest of the department to actually tap into this, this new data sets.

Josh: Yeah, yeah, it's it's a new kind of technology. And I don’t say that I don't want to be like this, hype, again. Oh, it's going to transform everything. We need to put it on everything. But it's, to say that is to say that the approach, like we were saying, needs to be different too. IT may or may not. You know, it really depends. And that's where it comes back to the people. You need to know your people, there could be people in IT that are like thinking about this and are thinking in the right way about this, but typically maybe not. And so that's it's it's not a SaaS, it's not a chat bot. It's fundamentally taking a business process analysis and applying like intelligence to it. Right. Now, the issue is that ultimately this is this is oversimplifying it.

Bill: Good. You know your audience, good.

Josh: Ultimately, AI is probability. It's probability stacked on top of each other with some cool equations. And so first of all, it doesn't have the responsibility of a person. And it doesn't have the just like intuition and the common sense. Right. So it's like a really smart intern that comes into your company and has no idea how things work. And so that's where you need to attach the right data. You need to put it within the right processes and kind of build structure that can hold its hand as it walks through things, you know. And so that's what it is. It's the ability. It has the ability to take action that's bounded by what you define, as opposed to a chat bot that you talk to. And you're like, hey, fix my sales department. And it says, hey, you know, you should lay off Jerry and then you should go and do this, but it doesn't actually create anything or actually output anything. So that's the that's the key. And it's a it's a new kind of way of thinking about it that has to happen.

Bill: No, I love that. And thank you for simplifying it. That's always good. And this is also why, we believe as, Josh and Aaron are two people on the team, we have another three here in the States, that we need to, take a different approach to this with the way we interact with clients. So, we're going to be in people's buildings. We have been, that’s part of our process, we are going to come to your facility, and we're going to engage your people, whether that's from an educational standpoint, whether that's building relationships with your IT team or whoever we need to do to get the data, whether it's sitting with those seats and really accumulating some of that data. We have some tools that can streamline that and we can use technology. But we also recognize that this isn't just a headline so that the CEO considers the board and say, yeah, we're doing AI. This is like, oh, here's some pilot projects we have. We’ve invested X, we received Y, we're anticipating additional return in the future. And not only are we starting on this process with these pilots, we're building a foundation for future expansion in that specific department or in other departments, because ultimately, to go back and what we said before, we do want to see transformational change, but that is going to take years, not because of the technology, but because of the people, process and tools that are available. And like the situation on the ground. And the adoption curves that are going to be necessary for, for employees and teams to embrace and internalize these processes so they can be that 20%, you know, that human in the loop part. And we can really continue to see learning, through that 80%. And hopefully 80 continues to creep up. But, how many years have we had SaaS and we still need human operators to deal with SaaS?

Josh: You’ve just gotta be anti-hype. You can’t get carried away.

Bill: That's right. So when we think about those small steps being in, people's businesses, talking with them directly, doing voice of like, seat, research, like interviewing people, looking at their data logs. That's, I geek out on that. So I get excited about the operational technicalities. You guys understand all the, the technology way better than I ever will. But when I look at that operational detail, and then we get the operational detail, and then we start to mesh it with these tools and these automations in this AI and this learning and the probabilistic thinking and those type of things, that's where we're going to see magic. And that's where we're seeing magic today. And when we start to see those use cases and we can’t guarantee this on every project. But when we start to see cash on cash return of 3 to 7 X in one year on a project, what ELT is not going to jump into opportunity for that type of return on investment, right? If they could do that on the stock market, they would buy that stock right now. So but if you can do that internally with your company, that's the path we're going on. And the other thing that excites me about AIS, is that we're not just coming at this from an experienced level of deep understanding of technology. We're bringing the operational experience that we have with our other companies that we own and operate and have owned and operated in the past, so that we're not just going to come in there and say, hey, you need AI, it's going to make you better. You know, just give us this stuff. We're going to help analyze what's happening on the ground and see those those situations. Take our database of knowledge of past implementations and engagements to say, hey, we've done this type of thing before. We feel this could be really right for you and help you get to where you need to be. And I think there's also a danger right now where there's a lot of folks who are spinning up ideas on how to implement AI and, God bless them, they’re these young entrepreneur types that are only AI, and they may have never worked in a business, they may have never sat in a boardroom meeting. They may have sat through a monthly or quarterly or annual financial statement review, or looked at key performance reports. And that that may not be needed 15 years from now in the AI dystopia. But today, that's still very relevant. And that's valuable, in the process. So I think we look at the, the operational intelligence that we bring to it and, and history and benchmarking. And then the innovative aspect of this technological shift and jam those two together if we can, kind of like what happened in the Industrial revolution, if we can take this new technology and apply to every layer of the process fanatically, that's where we're going to be humble enough to know that it's small steps, but then we can start stacking these 80% wins on top of each other. And that's where we're going to see compounded growth. And, that that's what excites me about, our relationship and what we're bringing to the market. As we close on this segment, maybe we need to talk about some real practical steps that we could give as advice for that CEO out there as far as like, okay, what are the next steps for me? Interesting ideas. I'm hearing the hype. I've got, I've got so much static in my brain about this, but I need to get down to the signal. If we follow the Steve Jobs approach to this. What do you guys think are some practical steps that we could, like a one, two, three, what a CEO should do today if they're thinking about something like this?

Aaron: So, two things. If a company has not adopted AI doesn't have Claude or Gemini whatever it is, I'd say identify those seats, identify those individuals who are gung ho about AI and just talk, have conversations with them, see where AI can be embedded with that and just have those conversations right. It starts with the small steps and identifying those people in the company that can make those improvements, because they’re there on the ground, they're getting their hands dirty. So start there. If the company hasn’t dealt with AI. And then the other one is if they're dealing with AI. This is very niche, which is what came to mind. And if we're talking practical, this is very practical. A few, I think, probably, maybe six months ago, everybody was saying prompt engineering is the field that you need to go into. It was you're always going to need prompt engineers like this is where AI is going to go. And now that's slowly kind of being like I don’t know if we need them as much anymore. And we still need them right now. But you can definitely see that we probably won't need them a whole bunch in the future is that's what I'm I've been reading. And I say all that there's a software called WhisperFlow, and for me, it goes Claude and then WhisperFlow. For those who know AI, that's a crazy kind of like stack. Well, WhisperFlow allows you to do is it gets rid of you don't have to type anything anymore. You click a button, you speak to it, and then you can click two other buttons or another button, and then it will rewrite your entire kind of brain dump into a specific prompt. And so it’ll re prompt it for you. Or you can have another shortcut where, say you're writing an email to an individual, you hit the key. You bring down the email, you hit another button and it polishes it for you. Right? That's where we're talking about the incremental change. You’re not having to type things out, and you can have it repolish what you just wrote and then proof it and send it. And you sent how many emails a day? A week, a month, right. That's compounding growth. So if you're if you have AI and you’re building AI, Whisper Flow is a is a software that I highly recommend.

Bill: Love it.

Josh: Yeah. So to kind of like sum that up and add just a little bit, I think that what Aaron talked about is super key of first of all, identify your people in your company and then you have to be using it too. You you know, you had you have to be getting your hands dirty and trying it out on your personal workflows, your personal lives. The one thing I will caution there, and this is the tech side of me coming up. Don't start creating ideas or pushing out things that you don't have good knowledge about. Good domain knowledge about, right? So we see this like all the time where it's these like, I think the Y Combinator guy, he like, made this web app and he was like, go check out my new SaaS that I just made. And people were just in the technical world were just like ripping it to shreds. And so, like, I think the humility thing is super important, just being like, okay, implement it, you know, implement it where you know that you can judge the outputs because that's that's the dangerous part. If you get to a point where you see an AI output and you're like, that's amazing, that's going to change the world, that's the hype. And oftentimes it comes because you don't have enough knowledge about the area that the output came from to be able to identify the holes, you know. So I think those two points, super valuable, identify your people, use it personally. And then just with that sort of caution, there, tacked on to the second point.

Bill: So the other thing I think that a CEO could think about today is, yeah, we need to they need to be using it. They need to find the people who are excited, kind of create a steering committee where everybody's talking about it. But then I also think getting outside strategic help may be the key. We started on this journey two years ago and it was very early. And the thing that got us step change is when we we custom consultation and then partnered with you guys, and E2M, and a couple other organizations to really hone our theory and our strategy around this space. And I was at a AI conference, very accomplished, head of AI, chief AI officer in a company. We're sitting and talking, and he said, Bill, no matter what you do, recognize that AI transformation is not going to happen in your company with your current team. You are going to have to either get outside help or create an entirely different department or company, because, and it’s nobody's fault. The people who are doing the current work today have to keep doing the current work. They don't have time to learn it and figure it out and start to implement it for you. And I thought a lot about that, and Ronik gave us some great advice on that. And we'll give a shout out to Ronik in a link in the, in the notes. But, we took that back. We didn't try and do it internally. We used third party resources, partnerships, those type of situations, and we're getting outsized results because we changed the way we thought about something that's new. And so I think if a CEO has to have a perspective because CEOs are not going to implement this. They're not going to be involved, but they need to have the right mindset around how they're going to help their organization grapple with AI and this transformation. I think it's get help, and it doesn't have to be huge dollars. But start with strategy, get some looks there, then maybe look at that data, start preparing the data for those moments. And whether that's organizing the data you currently have, or whether that's starting to capture data that you aren't currently capturing that exists, that's not being recorded and not being captured. And then you can really start to think with humility about de-risked pilots that can be done really quickly. Show good cash on cash return. And why does this matter for CEOs? One, they want to be in the game. Two they want to deal with the pressure they're experiencing because the board, the ownership group, whoever is pushing on them. Or maybe it's just an up swelling from the team where they're like, hey, we need you to lead us here. I’m seeing so many CEOs who are 50 to 65 years old who are retiring, and they're stating, I don't want to deal with this. I don't want to be part of this. And these are, these aren't CEOs who don't have resources and teams. These are multibillion dollar companies that they can allocate 100 million, $200 million to this problem. They just don’t want to deal with it because it's so different than what we've experienced heretofore. But, so those are some things that I think will be helpful and reflective for those CEOs to make those decisions. Well, guys, I'm excited about our partnership. I'm excited about where we're going. There's a lot of things, we’ll have some other team members on the pod later to talk about these things. We've got some great, partnerships with other companies, like we're building something here to go and change the market. So it's just a very exciting moment. And I just really appreciate you guys for being on today.

Aaron: Thanks for having us.