The Real Reason AI Fails

If you want to put it up on the website for people to read, you certainly can, but quite honestly, that's even dated because they're going to just be talking to the AI agent. Why would I even go read a Knowledge Base article? If I can just ask the problem I'm having in the AI can generate an exact scenario with the answer that I want. I can prompt the AI. Welcome to the show. Where AI takes the flow. Learage on the mind. Welcome to the AI guys. A podcast where we make artificial intelligence easy to understand for everyone. I'm your host, Lee Dixon. And as always, I'm joined by the man who knows no amount of AI can fix a broken workflow. Richeswire. How you doing today, man? I'm doing great. I'm doing great. This is our first recording of the new year, technically. It's freezing outside. It is in Florida. It's very cold here. But we have a guest who's in a much colder place than us. So we can't complain too much here. So our guest today is an engineer by trade who spent most of her career between business and technology helping organizations and teams turn complex technical ideas into practical actionable outcomes. She holds a master's in management of AI has completed formal training and data science in ML. And beyond her corporate work is deeply passionate about building inclusive tech communities. Maria Bessera, welcome to the pod. Hi, Lee. Hi, Rich. Thank you so much for the invite. I think you're a little too esteemed to be with us on something like this. Like everything felt so clean and concise. And sometimes our guest just likes to sail and talk about tech. This guy. So I really appreciate all the energy they're bringing in today. Oh, yeah. I'm always bringing energy. Whatever I go. Yeah. Well, I'd love to give a second just so you can tell a quick overview about what you're working on right now. I know you're super involved as I was saying in that intro around creating inclusive tech communities. So I want to give you a minute just in case you want to plug on some of the things you're working on in 2026. And then we'll jump into some of the things we're going to talk about today around AI. Sure. And again, thank you for the invite. Very excited to be here and talk about something I'm really passionate about, which is AI, of course. I've been working in AI before it was a cool thing. So I tried to do it. It was released. Yeah, exactly. So that's even more fascinating because we've been able to see a little bit of it before it became mainstream kind of thing. But yes, this year I've been working on with a lot of business units trying to help them move from ideation to actual execution and production, not only internally for their business units, but for the customers they serve. And as you mentioned outside of work, I'm really a deeply passionate about inclusive tech. And that's why I also ran an offer profit called the AI Ladies Lounge, where we host in person and online events for women trying to learn more about the worth of AI or women who are in the AI world and want to advance their skills. So yeah, very passionate about AI for what you can see. I love it. Yeah, I mean, obviously we've gotten to meet you a couple of times through the constellation world and everything like that. But constantly seeing things on LinkedIn about the different stuff you're launching and trying to really get out there and motivate people to get involved in such a unique tech. So yeah, appreciate you making time for Rich and Lil' O'Me to chat with you for a little bit. So I want to jump into it. I know we have a little bit of time today and we're going to go through some of the high level topics that I think you can really help on. And a lot of topics that Rich and I think have hit on in the past episodes, but really this concept around, maybe AI is a people problem in some ways, but that's actually kind of a good thing, right? So I think Rich shares a lot of the mentality in the sense that with things like AI, we have this ability to kind of get in its way or try to criticize or critique farther than we should as we wouldn't do with a human here. So I figured maybe we can start off talking a little bit more about why AI is more important about the people and the processes that it might be supporting in a scenario versus just the buzzwords, right? The models or what you're, you know, I'm using reasoning versus this mini, like really focusing on the core outcome of these types of positions. - Well, there are so many things we can talk about when we talk about the human element, but I would just start by saying that change is hard on its own. And when you add a technology on top of that that is changing not, and not monthly, but almost daily, it's hard for people to keep up with everything. I mean, this is even my day-to-day job, and even for me, it's hard to keep up with everything, let alone someone who has their plate full with a lot of tasks and activities, and then you ask them to work with something AI related that's gonna help them, like it's a lot for them to digest. So for people, it's hard to add that very quickly, especially with the how fast things are changing. So that's one, but also when we think about the technology, it's evolving fast, and the models are improving all the time. And it's important that there's a person that can help these organizations help them move from ideation to execution, because as I mentioned, it's changing so fast that sometimes you don't know what's new, or what else I can do for my processes or tasks. So yes, it's an element of change of mindset, and even for those who are ready to jump into the AI world, it's hard sometimes for them to grasp all the concepts to understand what's the possibilities or how I can even get it started. So there are a few things about training, but also the mindset that we can even talk more about. Yeah, we've experienced just working with companies across the world. I think there's just a lack of preparation to understand exactly how AI is going to permeate through the company, even culturally, right? Because, and this has kind of faded a little bit, but even the early days, and I guess even today in some channels, we always think of like AI is almost like it's cheating, right? Like if you're using Chatchy PT to write a proposal, or if you're using Chatchy PT to write your documentation, it's almost like you feel like you're cheating, right? Like if you're in high school and you're writing your term paper with it, right, you get kicked out of school. But I think it's such a magical technology that people just don't really understand or know how to really apply it in the way that they feel is the right way, right? Like, you know, it's versus like going from, you know, cheating, which would be one extreme, to being lazy, to being more effective. And sometimes it's one of those things where, I think mentally as humans, we always feel guilty when something feels like we didn't put the work in, right? Like if you build out a 10-page document or you write some code, everybody kind of feels like, oh man, that was really easy. So from a cultural perspective, I think people are, you know, having a hard time wrapping their head around, okay, what is exactly am I doing with this tool? And how does it actually, you know, what should I be, and it forces you to reimagine what your job is, right, and what your work is. And in our case, like when we're building out AI agents, we're basically telling people, you're no longer the person who's answering the support ticket, you're managing the agent's answering the support ticket or you're no longer really the lead coder, you're qualifying and evaluating the code that the AI writes, right? So, yeah. - More of a project management role. - Yeah, it's a mind shift. So I mean, I don't know what your thoughts on as far as like culturally, just people just wrapping their head around how it kind of permeates through their work. And maybe that kind of that nervousness of, of will A, how do I adapt and B, you know, obviously, maybe not so much anymore, but people always have this fear of it replacing them. - Again, I think it goes back to their minds that I was mentioning because teacher example, today we use Excel to do calculations and we use sometimes a calculator to get to certain formulas and results. And we are not doing everything by hand. And if we go even farther, we are not creating fire with sticks and just out in the wild, right? So it doesn't make us lazy, but it makes us more effective and more productive. And it means that we can accomplish much more than what we could in the past. So I believe it's the same with nowadays with this technology. You have the tools and the options as your disposal. You should consider using them because that will equip you with more time, with more energy to focus your attention on a strategy or on other tasks that perhaps you haven't had the time to focus because you are so busy with those mental processes. So I believe it's amazing time that we are living. And to your other point, like the way for us to feel a little bit more comfortable is by experiencing these tools, by trying them, by not only reading, but actually building the agents or at least trying them with local tools or no-code tools like Raya. So that will actually show people the art of the possibility. And once you see it, there's no way back. Like you want to keep building and doing more and improving the way you work. So I guess it's just a matter of getting yourself out there and building that first agent. And then you'll see the beauty of it. - Yeah, I just feel like this one shift that's taken place from like a year ago, right? Is like this concept around like outcome here is really just based on adoption at this point. Like if you're not trying to at least understand it or apply it in somewhere within a business or within just your daily life, like that is where the biggest lever, I think is to pull right now. If you're having team members that are pushing back against it or just, I really don't think it's gonna help me here. like you've kind of. have a dead wheel on the car, right? And it's going to make it a lot harder to move. And there's another point you mentioned about the fear that some people have about replacement. I unless your job is a full-time copy and paste information, I don't think AI will replace you. Guilty as charged. AI is not deterministic as we know, right? So we still need the human in the loop. We still need our sub-diet-matter expertise to validate outcomes, especially if we are in customer-facing environments or building customer-facing products. So we still need that human element to help us validate. Yeah, I agree. Yeah, I mean, it's that pathway that I don't think people have really, there's not really great resources or really any great guidelines, especially for companies that are going through the evolution of AI, of how you would advance your workforce. You know, we're working with one CSI company right now and they're really forward-thinking. It's a company called CrossCAP. And part of what they're building out right now is they're basically creating a mandate for all their employees that by the end of the year, they'll have each employee will have three agents working for them. And so it's such an interesting way to put a mindset, mind shift on your team and say, okay, listen, salesperson, you're going to need a BDR and SDR, a proposal writer, whatever you need to enhance your job. I want you to build those agents with somebody. Obviously, they're not necessarily going to do all the work, but they have to be the, they understand the domain expertise. They understand the process that needs to happen, right? They don't necessarily have to understand how to code with AI or even build an AI agent, but what they do need to know is, hey, I know that half my day I'm following up with leads. So if I can have an AI agent do that. So really, it's this onus of like somebody from the top down has to say, listen, I want to have an agentic workforce and I want to motivate my people, my team to allocate work to that workforce. And now they're becoming pseudo-managers. So like you're saying, like, it's this combination of human in the loop because you're managing these people or these agents. But you're also continuing to do your job, but you're focusing on those things that really only a human can do, right? And I I thought that was a very interesting way to approach it. What are your thoughts on that? Do you think that's going to be kind of a trend? And is that kind of what you're seeing out there? Or is it still kind of early? I wanted to say something, sorry, that they've been into something you mentioned and into the process part and subject matter experts or people who are in the business know their process in and out. Like we may know about AI, but they are the ones who know their task, right? And sometimes we ask them, how can AI help you? But they sometimes don't even know because they are so used to doing their work in certain way that they don't know where to start. So something that I've been talking with businesses is to start mapping their processes. Because sometimes you know everything by heart, but when you see a written or a drone, you can actually go into every task, every step and see how different agents can connect and then create these multi-agents that then you can supervise. I believe that that company that you are mentioning will end up with many more agents than three per person because once you identify one, it's going to be much easier to go and identify the other ones. And I think that's something that I've started to see more to your question about the trend. I've started to see more companies now focusing more on the holistic process rather than in one task. And that's when I help accelerate that adoption. Because in the past, we were focused on, okay, in this task, I'm going to automate it, I'm going to use AI. But when you start looking at it from a high level perspective and seeing how all your processes connect, not only for you, but for even your team or other departments, then you can start thinking about how different agents can work on your task, on your team, on your departments and your company. And then we'll end up with more agents, more than three per person. I'm sure you'll see them. And I guess to your point, evolving from kind of what I would consider, like you mentioned, task-oriented AI to more process-driven AI. And that kind of parallels the evolution from tool bit tools. Most people, especially last year, are primarily when they say we're using AI, they're probably referring more likely to tools, like chat GVT or co-pilot or get or cursor or whatever they're using it for. Whereas the evolution of tools to agents now opens that opportunity. You're going from something that's tethered to your desktop and you still need a human to work the tool. Versus, okay, I want to make an autonomous and untether it from my desktop. Now you actually have, now you've opened up the world too. Well, I can actually do full workflows, full processes. And I like what you said, I do think it's one of the most critical components of building out an agentic workforce. And also, just in general, getting levered out of AI is of the three big pillars of AI, one is being workflow, right? Not at many people, as you mentioned, even have it written down. People know it because it's in their brain and it's just kind of habit, right? They go to work and they do the things they do. And they've evolved. And it's ever changing. And sometimes the other processes depend on a variety of variables that nobody really knows except for this person or that person. But like workflows is one of those things where as you start to move things to an autonomous AI or an agentic workforce, you have to write it down because you have to figure out what part of it you're going to be able to automate. And the other two pillars, which I'd love to get your minds, your feedback on, at least my opinion, is the other pillar is the knowledge base, right? The data of the business, right? All the data, whether it be customer information and knowledge base around your products, services, your sales, your marketing, your training, like think of it like all the information that's proprietary and specific to your business, that's another area where we haven't really done a good job of writing it down. Maybe we have a knowledge base, but maybe it's a little bit dated, you know, because until now nobody really cared about it because they're like, yeah, we have a knowledge base and we updated every once in a while, but nobody ever goes there, so nobody really cares, but now it's like it moves to the front of the line, right? Now it's more critical than ever. Yeah, a lot of these businesses now. Yeah, and then, you know, so I love to hear your thoughts on, you know, workflow and knowledge base and kind of what you're seeing and the process people are taking. Sure, and I'm actually working with one of our important business units in Jonas, and they are trying to come up with a customer support agents, but of course they can not yet, just because their data is not up to date. So we see this every day, right? And even more importantly, they may not have the data updated in all the different departments, but they don't have a full standard process on how to distribute the data to everyone coming from the same data source. So we found that different departments were creating their own documentation. Sometimes there were inconsistencies among departments. So I was like, okay, first of all, we need to standardize the process. And that was our first step. Going department by department understanding where people was getting the data from what do they do with the data and what format and so on, because everyone was doing it manually as well. So I was like, okay, we can use AI to start with these process and accelerate the documentation piece. We don't have to wait to build the agent. We can start using AI right now. So they've been using AI to get the source data and transform it in different ways that can be consumed by the learning and development team, by customer support, by the training team, and so on. Now that they have the data ready, we're going to start working on the customer support agent. But to their point, yes, it's very important that we have a processing place, then our documentation up to date, and then we can go and build the agent. But it doesn't mean that we have to wait until the customer support agent to use AI. AI can accelerate so much the documentation piece, especially if your business haven't updated it in a while. And there is that simple trick that I keep telling people, you can even use the station in your word document and say how the process works like. And just say it with your words and then pass it through an AI assistant and ask it to refine it into the format that you have. Now if you are going to do these ones in a while, that's fine. But when you have to scale, then you need something like, again, like Ray, to go and build it automatically and that it can help you and keep you posted when something a new documentation is ready to be processed. So there are many different ways, but it goes back to process data and then the agents and all this process, the change management piece, right? Because we cannot just mandate to use a tool if we have not trained people, if we have not told them, this is how this is going to look like if we have an involved people in the process design and so on. So there's a key factor that goes into involving the right people at the right time from the beginning, basically. Yeah. There's one thing that I want to touch on that you just brought up to. I really feel like where we've now shifted is away from maybe the 2024 or 2025 of like, let's do a presentation on how we're going to use AI. You can't just get up there anymore and be like, you know, this is what we're going to do this year. I promise you really have to start either just taking swings and seeing what works and what doesn't. But I also think from some of you you said a little bit earlier is that it's really hard to sell sometimes inside of an organization, pick somebody who, and say, you're going to be our AI person now because they do have day jobs. They have other responsibilities and it really going to take your mind off of that and focus on this solely even if you give it 50 percent, like you think that would be a lot, but like sometimes it's actually more of a detriment to the process than a benefit, right? It's not having some dedicated resource, even if it's external, be the driving force to make sure that this is getting put into production the way that you want it. Oh, yes, for sure. That's super important. If there's not a person responsible, then you're going to be the danger projects. That's something that I keep telling people, if you define a project and idea, you need a driver that takes it from ideation, way to execution, that has weekly stand-ups or even daily stand-ups, depending on how fast you want to go. But someone who is really responsible for taking care of those projects, that's super important. Otherwise, it's just going to be one work project. I think all of this, when you really kind of tallied up, is the reason why it is so difficult to launch sophisticated AI solutions because it's a trickery that AI has created because it feels so simple. We load up ChatGbT and we start talking to it and it just seems so magical, which is what obviously brings and aloers people into the process. But when then you actually say, "All right, great. I see the power of this thing. I've experienced. I'm able to chat with this amazing large language model that's been trained on the world." Now, how do I translate that power into my business? Then basically, what you realize is, "Oh, man, I have to train the AI on my business because the generalized models don't understand my business. Don't have access to my data. Don't understand my processes the way I would like to have them." Now, that's a non-zero amount of work, right? Unless you're like a pure startup where you built out and have your data is clean and you relatively have good access. If you're dealing with a 30-year-old software company, like we do on a daily basis, then you're dealing with nightmarish scenarios where you have documents for 20 years ago that are in God knows what format. Then the second thing is, "Oh, man, what do I have this agent do?" Now, you have processes that you have to work flows that you have to sit here and work on. I'm putting somebody on top of this project and you might even have to, if you have third-party applications, like you have a CRM or a support system or whatever, then you have to start thinking about, "Okay, do I want to give the AI access? Do I have to now start if it's a larger company? Maybe I need a created data lake." When you start really combining all this, before you even get to AI agent number one, there's a tremendous amount of complexity and work for, I would call for a mid-mid-to-large company, especially if you're wanting to do some deploy AI agents that actually are going to really drive a lot of ROI, right? Sales support ops. I guess one of the things that I'd love to hear your opinion on is when you go into a business, just pick a, like within CSI, just pick a random BU, and you come in first day and they're looking at you saying, "I need you to help deploying our AI." What is your checklist in your mind that you go through? Who do you talk to? What questions do you ask at a high level to understand, "Okay, is this going to be, how prepared, how AI prepared is this business versus, one of the some red flags that you typically see, it might be interesting for people to do their own evaluation, our own audit, and to a lot of it, are we ready for AI? Because a lot of people, I think they just jump the gun and they hire some random company or they buy some random products, thinking that's the solution, the reality is. Yeah, a lot of AI it literacy and they also have low business understanding outside of how to operate, right? They don't know if they take it back and were to rebuild it all today, how they would go about doing it systematically because they just did it on the fly until it got to terminal velocity. Yeah, so what are your red flags? How do you evaluate in your gut? The first question I ask is why, why did you want to build a customer support agent? And it may seem obvious for a lot of people to have one and it makes sense for them, but I run into a company, believe it, or know that told me, "Oh, we just received like five customer support tickets a week and we just want to create a customer support agent because our competitors have one." But in reality, it didn't make sense for them to invest in an AI tool when they were not receiving enough tickets. So, that's the first question that I would ask for any project regardless whether it's a customer support agent or any AI project at all. We should understand what's the rationale behind it. It said because of pressure, the market, or because we really needed or our customers needed. And then going back to what we were discussing about data, data readiness, what data do you have? Because that's another surprising answer that I got once when I said, "Okay, you want to build a customer support agent? What data do you have?" And they're like, "Oh, we don't have any data. We don't have any documentation." So, that's the second question, checking out your data, where the data is coming from. It said, "Abtodate, how often is it maintained?" And so on. Another key question to understand is the process, how is it today? Is there a process in place? How, who produces the data? Where does it go? And so on. The other aspect to understand is who is going to be responsible for the project? Because sometimes I go as a kind of consultant advisor to our business unit. Sometimes I help build the proof of concept. Sometimes I stay with them until the end. So if I'm not there, then there should be one person as we were discussing that is responsible for the holduration of the process. We also need champions or sponsors. Like, is it someone sponsoring these? Very important to understand. Did you have technical resources in house? Because that will also help determine whether you want to build something in house or whether you want to buy enough the shelf tool. If you don't have resources, it's going to be hard. So that decision of buy versus build is very important to bring it up early in the conversation. And then what's the need to go to the market in terms of the speed that you want to go tomorrow in a month? Because that will also determine how many people we need or how many or what resources we are going to have to allocate. So those are the key questions. And red flags, not having data, not having one person appointed to the project. Just doing it because my manager told me I have to do it. Those are the main red flags that I would mention. Yeah, gotta have commitment. Yeah. Exactly. But you also work in a lot of customer support project. What's all the red flags have you seen in these AI projects? I think, you know, for us, when you think about the big risk factors of deploying AI, you know, although our job and I kind of what we try to do with our platform is to reduce those risk factors, right? Like to take a lot of the complexity out of it. But when you're talking specifically about a business, one big red flag, I always, we always kind of look for. And it's a little bit similar to what your your question about why is, and it's typically around, they might have an idea of what they want to do with AI. And that's their first idea. And then we try to like evaluate like, okay, is that going to it might just think they might just be thinking, oh, this would be a really cool idea to do this, you know, but not necessarily evaluating ROI. So the big for us, because we have to deliver ROI, a big red flag is as an undetermined outcome that you can actually put real dollars behind. So if they say, I want to, I want to build an agent that does X, Y, and Z. And then we would say, okay, well, how would you, what does that equate to in the sense of human hours? And if they say, well, if you're able to do this, then it might, it'll save us five hours a day. And then I'll be, okay, then that's a clear pathway. But if they go, oh, there is example of five tickets a week and just be like, I just don't want to answer him. It's like, you're not going to spend the money to do that. Yeah. So, so yeah, things like, you know, ROI is another, you know, kind of a similar red flag, but it's a, it's kind of similar, it's a kind of a similar to the Y question you had. Another one is the systems that they use is a huge problem. So integration is like one of the biggest challenges we have with. And it's not a challenging thing in the sense of like technically. It's a challenging thing because A, you have security compliance, B, you have numerous systems that are, in some cases, legacy that are hard to integrate with, not, not necessarily programmatically, but just hard to integrate with because it doesn't necessarily align with what the agent's doing. When you're talking about an AI first solution, you're trying to put as much energy into the AI as you can because that's the magic, right? But it's also like, as you mentioned earlier in the, in the pod, it's a non-deterministic system. And so when you start to pair that up with deterministic systems, like Salesforce or HubSpot or Zendesk or whatever, there are some changes that have to happen sometimes in how you kind of evaluate and how you integrate. But building those workflows, like workflows are not just about if then, then do this, but it's also about go to this system, grab that, go to that system, update this. And so getting integration set up and running, I'm shocked actually on how few businesses, especially within our universe at CSI, have actually done, and they have hardly any of them integrated their core systems with anything. So sometimes it's a brand new thing, right? So a big question I would ask is, where's all your, like you said, where's the data, but more importantly, like what systems are you using that need to be updated, right, your CRM or whatever, and have you ever done integration with those systems? Because sometimes they go, no. And then you can assume that there's going to be some back and forth with regards to access, control, and stuff like that. There's also going to be a rigidity though on that right, Rich. I mean, like one of the some things that we've seen is someone gets to the point where it's doing what you want it to do. And it's like, but what about in this very niche fringe scenario when this happens? Like we need the AI to do this. And you're like, do you though? Like, you know, what does a human do right now? And they're like, I don't know what the human does, but yeah, I have to do it. And you're just like, okay. We definitely fall into a trap where people try to make, they don't embrace the full minds of AI first. Right. And they try to hold onto their human behavior. That is a thing that I, you know, that's not something you would actually determine on your first day asking them questions, but as you work with a company, what you'll start to figure out is they keep trying to force fit human oriented decision making or human oriented process into the AI. And it's not exactly the best use of AI, right? It's not like how it should be used, right? It's almost like they distrust it or they don't want it to have full autonomy. But the reality is, is that you're going to have to kind of at some point trust and let it go. I mean, we always joke about it. But like when we deploy customer support agents, you know, we'll have people when they're testing it, they'll say, well, it got this, it got this kind of wrong. It said this, but it should say that. And I'm sitting here going, do you read every single message that every single support rep that you have rights? Because I can guarantee you that I'm sure that, you know, like you're being very, very picky. You're holding AI to a much, much, much higher standard because it wasn't the come through five minutes before you shut down for the day. Like those are getting a short answer or not getting answer. You're telling me that you're, you're telling me that all of your support reps that, you know, are answering to exactly how you wanted to answer. So I do think there's like, they put on a different hat and the expectations are high. I think, I think that's an interesting challenge. I think it's a challenge that very few people really truly understand, which is everybody, because it's AI, everybody wants it to be their own. They want it to be custom to their business. They want it to speak and have the tone and the cadence and the mindset like they, they, it's like a duplicate of them, right? It's, it's a, it's a mirror image of their ego, whatever role they put, like be plain. And you just don't get that with software, right? The software doesn't have a personality software doesn't talk to you, but now we've, we've entered into this new realm where the, the expectation of how people want AI to behave is, is, is extraordinarily high. And that's why I think a lot of projects fail because to reach that high efficacy, that high bar that people are expecting, you have to have a really sophisticated system and you also have to have a platform and you just have to have a lot of pieces in place to get there. And that's, I think that's the big red flag is setting, setting expectations, understanding what they're, what they believe is to be success, just like any other project. What the ultimate outcome that ties to ROI and what kind of integration needs to happen in order to make that AI autonomous because if there's any sort of human and there's one, there's a difference between human and the loop, which is I'm overseeing, I'm checking it, I'm maybe doing some training. And like human dependence, you know, yeah, human dependence, which, it's fine. And some processes, you have to hand it off to a human and maybe they have to check a box and it moves on. But you want to try to suck all the manual labor out of it. Otherwise, you're always going to have that inhibited, you know, in your workflow. And I want to talk a little bit about something to mention, Rich, and it's the idea of having that AI mindset about our processes because sometimes what I've observed is that a few people are trying to just have an AI do what a human does. But it is preventing us from thinking or imagining how processes should be working or how they are going to work in the future. Like today, for example, maybe there's a person generating a report sending a via email and then another person reads it. Let's say that's how a business operates. Then they have an AI doing those steps and then the human kids review on it. But what if we can have everyone get access to the data in natural language, like we can do that today or things like transcript of meetings, like, yes, maybe we can automate that transcripts and the transcript to everyone with a summary and everything. But what if we don't even need meetings? So the idea of having AI just automate or improve every process just as we have it today, I feel it's preventing us from thinking or imagining it, how these processes can be optimized for the future. That's something that I would encourage everyone to start thinking of. How your processes are going to work in three, five years, maybe even in two years, because with the speed of AI, there are so many things that we can start accomplishing and improving the way we work and connect with other teams and departments and with our customers more specifically. Yeah. I mean, a great example to kind of extrapolate on one of your examples is, let's say you have a scenario where you send your customers a report, right? So we actually do this internally. We have an AI agent that generates reports for our customers on a weekly basis. But maybe you know, this is what humans used to do, right? We built out a report and we're going to, and we're just, we're trying to get the AI to generate that report exactly how we used to do it because we're creatures of habit. But to your point, why not just get rid of the report and send an email to the customer and say, hey, ask me, you can ask me anything and you can generate any report you want, right? Because it has access to the data. It's smart enough to generate. So some customers might might might look at that standard report and go, God, this is a really crappy report. It's not really telling me what I want. And some other customers might like that report, but you can basically go in there and say, well, yeah, I'd like to know this, this, and this, and this. And then the AI just spits it out. So you know, it is a, it is a mind shift. You know, it's, it's getting away from when we think about productivity and we think about the output of humans, typically the output of humans are measured in some sort of hardened output, especially in the, in the knowledge workforce, right? Which is what we're basically talking about today, right? So I guess in technically in the blue collar as well, right? Like, I go make a car. There's a car. But in the knowledge workforce, I make a document. I make a report, right? I make a proposal. I solve a ticket. So there's always some sort of, you know, document or some sort of proof of work, right? And when we move into this more fluid arena of AI agents that can, at any point in time, regenerate any sort of information from, from the history of how long it's been trained on, and, and dynamically created on the spot within seconds, it does, like your point is, is dead on it. It completely changes how, how businesses can operate, right? Like, I've great examples, you know, I, just for kicks, like we were at a, we were doing a sales presentation, and you know, for a few hours before the presentation, we, I was like, I'm going to take my sales presentation and I'm going to pump it through AI and have it speak to the audience with a metaphor that they would understand. So they were like in the oil and gas industry. And so I just generated a sales rep presentation that talked and uses metaphors in the oil and gas world. And I was just showing it as an experiment. But, you know, there was no like, I didn't have to let go and create a lot. I just took exactly the same message, the same. You know, I mean, obviously I had to go give the pitch. So I was stumbling through kind of because I don't, the AI generated it. It's almost like somebody else built it for me and I just had to learn it. But it's a different mindset, right? It's like I can, I can now dynamically create and shift. I even, with customer support rep or bots or agents that we've built out, one of the thing to your point earlier when you talked about the company that had their knowledge base was a mess and you started talking about, well, you can use AI to generate that. This is exactly what we tell people is like once the AI is trained and up to speed, you can have it generate the entire knowledge base completely different in any language, in any format in a much better way. And so, and still people don't quite understand it or they don't quite believe it, right? They're still going, no, no, no. He hated, he hated its format for a decade, but you just want to keep going back to that. Right, right. How do we make this better for everyone involved? Yeah, it's like you don't ever have to write another knowledge base article. Once that ticket is solved, we can have an AI agent grab that ticket, grab the resolution generated knowledge base article. And if you want to put it up on the website for people to read, you certainly can, but quite honestly, that's even dated because they're going to just be talking to the AI agent. Why would I, why would I even go read a knowledge base article? If I can just ask the problem I'm having in the AI can generate an exact, exact scenario with the, the answer that I want, I can prompt the AI. So, even just static content is, is kind of an outdated concept. Like, why, if I can generate from an agent or from AI, why do I need, why do I need the static content? I think that's obviously what Chatchee BT kind of showed everybody in the world, right? Like, if I can ask Chatchee BT about how to make, you know, a specific spaghetti with certain ingredients and it can dynamically generate a recipe, then why do we need recipes.com? Right? And, and it's, you know, I don't know what's going to happen. I've always been on the court of the AI is, AI is going to generate a much bigger internet, but then over the last six months based on kind of my experience, I'm like, maybe it's going to shrink the internet because in the sense of life, I'm going to be able to do that. like if it's if I can just generate 1000s of recipe websites when you've got one right maybe websites. It's got it all. I mean, completely not this top not a topic for this, but maybe a segue for the next episode. We have the next episode. But next time we have Maria, we'll talk about the death of websites. You're right. Yeah. Maria, we appreciate you coming on. I know there's so much more we can cover, but obviously we'll have to have you back. So, you know, thank you again for coming on this episode and sharing your insight with us. And my pleasure, it went by so quickly. Thank you so much for having me. Yeah, absolutely. And for everyone listening, thanks again for joining us on this episode. If you haven't already, hit that subscribe button to stay in touch. Also check out our substack for more content, which will be linked at this episode to continue this discussion. Until next time, stay curious, stay informed, and we'll see you on the next episode. Take care, everyone. ♪ China's every week ♪