Stuff like that costs a hundred thousand dollars with AI. We can move so much faster and so much affordable and so much more customizable. That's a very different game.
SPEAKER_02How would you describe this concept or this paradigm of vibe coding to the listeners?
SPEAKER_00When I talk about vibe coding, it's really just about creation through your words in the application or any type of AI tool you're using.
SPEAKER_02A client says, hey, I'd really love this. Where did it go from that, hey, I've got this idea? to what did it look like for you to be able to say, here's your idea, it's now working.
SPEAKER_00Use a framework called the Scouts, a framework that I've created, scope the objective, and Doctor East or outline a plan, unleash the builds when we go into testing and optimizing. I do think there's a room for a lot of people to build their own personal tools.
SPEAKER_02Eaton Pullinger is a leading AI consultant and founder of Makesense.ai, known for transforming mid-sized businesses with no code AI solutions. From marketing routes to building powerful custom applications, Eaton empowers non-technical teams to create, automate, and scale fast.
SPEAKER_01Welcome to Using AI at Work. I'm your host, Chris Dagle. Each week we'll be learning how today's business owners, entrepreneurs, and ambitious professionals are getting more done with smart use of tomorrow's tech. Let's get started.
SPEAKER_02Right now, every business leader is asking the same question. What are we going to do about AI? If this is you, ChiefAIOfficer.com has the answer. We give you a simple path forward where we provide executive and team training so your people know exactly how to safely use generative AI in their day-to-day. We also manage the deployment and implementation to make sure tools actually get adopted and deliver results. And we'll also guide company-wide transformation so AI becomes part of your operating system, not just another shiny object. The companies that act now will increase productivity, cut costs, and grow faster than their competitors. Those that wait will get left behind. So if you want to make AI work in your business, visit chiefaiofficer.com and see how we're helping companies of all sizes finally get results from AI. Welcome, dear listeners, to another episode of Using AI at work. My name is Chris Dagel. I'm the host of the show. And today our guest is Eton Pollinger. He's a friend and the owner of Pollinger.ai. And today's topic is something I'm very excited about. And I'm sure if you know anything about the topic, you're going to be excited about it. The topic is, I hate to say it, but it's, we'll call it vibe coding. Although I think after today's conversation, you're going to understand it at a whole different level, and you're going to realize that it's way more than a tech bro activity, that it's something that even as an executive, uh, you can create some powerful resources for your company, uh, for your clients. You can create uh products to sell, all kinds of stuff. Uh simply using natural language with uh today's technologies. Eton, before we get started, uh maybe share a little bit about what you've been up to, particularly in the past, say, six months of really digging in on this stuff.
SPEAKER_00Yeah. Well, and uh, Chris, I am I know we've been trying to do this for probably a year, so I'm very excited to be here, which is funny because we see each other so often. Yeah. Uh, but I'm so happy that I'm here. So um that's one. Two, uh, what I've been doing. Uh yeah, so working on building end-to-end solutions with AI from automations, agents, applications with enterprise clients, um, which is, as you know, surprising because I never did that before. AI. Uh it was from the marketing world. Uh, but that's kind of you know what I've been doing. So taking, you know, figuring out different solutions and using AI to solve them and creating interfaces for people to interact with those solutions. So um it's been kind of you know a good ride. That's the last six months have been crazy, as you know. So I imagine that's the window that uh yeah.
SPEAKER_02So for people that don't necessarily maybe they've heard the term but they don't quite understand it. And I know that that vibe coding is is kind of like just a generic term and that what we're gonna talk about today is actually much more robust than that. But how would you describe this concept or this paradigm of vibe coding to the listeners?
SPEAKER_00Yeah, I mean, you know what's funny is I think at first when I heard vibe coding, it seemed like such an appropriate, oh, like a fun little term. You're kind of feeling, you know, you're you're coding with what you feel, and you're kind of you don't really know. Um, but now I honestly look at it as a I mean, what a powerful expression from our mind to creating something. And I don't even know the right word for it yet, but it's really learning the right, you know, learning the what the right ways, of course, to do this. You any idea starts to come to life. So that is a uh obviously a very different you know conversation. So when I talk about vibe coding, it's really just about you know creation through your words in the application or any type of you know AI tool you're using. But I think it's really that um, you know, I hate to get, you know, I heard Alex Ormozy say this before. Um I won't quote all of it, but he said something about the the the more you are more powerful, the quicker you can go from thought to creation. Uh and when I look at vibe coding, so to speak, it's really just a framework of doing that. Coding has always been a work of creation. Now you can use your words to create. I mean, it's really powerful. So that's kind of how I feel about it, or you know, this playground of it. It's kind of yielding this power.
SPEAKER_02So, listeners, when he says going from you know, words to coding, that's exactly what he means. Typing in, hey, I'd like an app that could do XYZ. And behind the scenes, AI takes over and starts to code, like like computer code an app based on your, you know, your vibe, your idea. And we're going to talk about some uh pretty sophisticated uses of it today, but I want you to know that that is out there with tools uh that we'll discuss today, and I'll share some of my favorites. I've been vibe coding, I must admit, I'm blown away with what I've been able to create for our clients, for our internal teams, simply by saying, hey, I got this idea. This would be cool if an app could do X. Pretty incredible. So, Ethan, first off, what tools are you using? And then what would you recommend others use? Because I know that you've gone through a lot, but you're doing some pretty advanced stuff.
SPEAKER_00Fair. Um, I use a wide variety. But if I had to choose one in my current stack, it would probably be Replet, because it can get you so close. I mean, so close to something so powerful pretty fast. Um, especially in their newer, you know, they've had some new releases that are just unbelievable. I I think you know the tools are great and there's so many. I mean, truly, if I was to say, if I had to choose again, if I had one for the building element, for talking applications, not automations necessarily, even though they have an automation, but I wouldn't use it for that. You know, I think to answer this fully, I would have to give you the different kind of use cases I see when building solutions. And I think that's you know, uh a quick rundown would be, you know, essentially we we have the whole, you know, just typical kind of chatbot conversation. You're talking to a smart agent that can help you a co-pilot of sorts that knows custom GPT or whatever. Right. I think the next evolution is, you know, or going up is to an agent where now it can actually go and interact with different things for you. You're still operating it, but it can go and do, you know, hey, help me with my email, go do that, right? You know, maybe you've implement some type of rag system. It has knowledge, information, it you know, it has its own knowledge base. Uh, moving into kind of more of an automation type project where you're just you can have agents in an automation, but the trigger isn't necessarily you saying something, right? It can happen from an email coming in, a calendar came up, an event came out, and a full automation happens. You you maybe you trigger it, but then a whole sequence of events happens that you do not need to touch afterwards, moving into maybe some type of dashboard project where a lot of data comes into place, it displays it, then can do things for you with it, and then to you know, some type of mini app, uh a minimum viable product or a full application, you know, is the way that I see these different projects as they're coming. And I'm sure there's more nuances, but just to describe, so each tool, okay, so to speak, there's a tool that works best for each one of these things or tools that's comprised of them, right? Um so if I were to say research, you know, which is a big part of my method when I'm building stuff, I'm all the time researching and reading in the best practices and documentation libraries, and I'm having, yeah, I do all the heavy lifting, right? Uh but some of the research tools might be perplexity or might be even menace. Go and find a lot of different things from me and creates different reports that I wanted to while I'm building so I can have the next thing worked out. If we're talking chatbots, you could start off a custom GPT and you know and work with that. You could use something like pickaxe, which I enjoy to use, you know, to for certain projects. So I know it's it I'm going from this, you know, the the the stack is expanding a bit here, but if I were to think about it in categories, you know, there's really the main LLM kind of choice of where you want to be. Uh then there's the overall orchestration tool, the one that does the automations that one might want to choose. The one that I'm using the most often is in Aiden, uh, because I believe it's you know flexible enough. And then there's also the other tool of this, you know, where do you want to build these prototypes? I like to build my prototypes in Replet. I do use Windsurf with Claude Code for more complicated uh things as a whole, or overall, you know, I may build a prototype in there and then I take it into making more complicated things in Replet. In uh sorry, in Claude Code and WordSurf. So that was a kind of a long answer. I hope that kind of brings us somewhere uh that relates to answering that.
SPEAKER_02And just from what I know of the space, that's like a very small sliver of the available tools that are out there to do this this kind of natural language interaction that results in an app or a website being built.
SPEAKER_00Oh, there's so many. I mean, I again there's so much opportunity, so many different ones. And I think really the main stuff isn't necessarily the main event apps, like say Replet, Lovable, whatever people are doing to build their prototypes in and some apps. But it's actually the um the things you can connect to. And that's where all these stuff, like the uh model context protocols, the different tools you can connect to, um, you know, I have a whole lot more of those than I do the main apps. Because it's kind of like a all these different superpowers I can enable these apps to do, and I just need to know how to connect them together. And it becomes suddenly a beast.
SPEAKER_02So let's let's do this. Walk me through uh a recent, like a client says, Hey, I'd really love this, and what you where did it go from that, hey, I've got this idea to what did it look like for you to be able to say, here's your idea, it's now working.
SPEAKER_00Yeah, I mean, I guess the the one that comes to mind, you know, since I've been working on it quite often, but it's it's there is um we'll talk, you know, I can bring up this uh sales training uh environment, you know, that someone wanted to train their sales team. They were bringing on 30 new salespeople uh that they needed to train in about eight weeks. Uh this is when we start a conversation. Um and they were just kind of thinking about what would be the best you know uh way to do that. They've had an idea of some type of uh bot that's trained on all their knowledge because they also train people and they have a whole lot of knowledge. Um, you know, this individual has over half a million followers for sales training type stuff uh online. Um and uh essentially what went from there is that I I use a framework called the Scouts uh framework that I've created and developed over all these solutions that I've been doing. Uh for those uh listening, yes, I was in the military, so the name is uh not just cutesy, it's a it has a you know it has some meaning uh besides the boy scouts that you may think of here. Um but the the the way it goes is actually this is how I also think is I'm very objective. You know, I have a what's the objective? What are we trying to do here? Is always the first frame that I have. And then the next thing is how do we break it down into actually getting to this goal, right? As a as any goal-oriented individual knows. So the first step of scout is scope the objective. So we figure, you know, so I figured out, okay, so what do you need? You need to train people, good, okay. We need to score, we need to also understand how they're performing. Are they actually using the platform? We need to track that so we can know, right? So the overall objective is to be able to train people, reduce the uh the work and overhead that's required because there's not enough team to train all these people, right? Uh, and also, so once they're on payroll, they're they're on payroll, right? You gotta get going. All these people are there. So our objective was to create a environment that would help train his salespeople, as well as give performance reviews and be able to track 7 30, 60, 90 a day, how often are they using it, what are their actual things, and on top of that, using all of his intellectual IP. So it goes from really just breaking down what are the components that come through our objective. We want to train our salespeople, great. What does it mean? We break it down to all these other components, right? From there, it's about researching, you know, what else is like there. There's gong, there's you know, all these other tools that are doing this, right? There's you don't have to necessarily always invent things, but you can borrow ideas. So looking through what's there, what's worked, what didn't work, right? Also important to see what doesn't work, right? Um so going through a research phase, and that's where we have in our scouts, right? The the C is conduct research or conduct recon if you want to follow the fun game, right? So you do some recon, you collect all the information, competitors, people who are doing stuff in this space, all of that type of jazz, using that then to move on to the next step, right? So in this scouts method, then we conducted our research. We need to outline the plan, we outlined the steps that the the build-out we create, you know. So this is where um this topic of just enough knowledge, right, and and the right engineering words, and you can really get far. And this is where we do our PRD, our you know, our product requirements uh document, what needs to be in it, what doesn't. We use stuff called you know user stories, which every product manager knows, most of us don't think about, but we don't have to know how to do that, right? We just need to know the word, user stories, which shows us when I, you know, what user stories basically, as a user, I do X, I want a Y to happen. Very basic stuff, but it really helps AI, right? So if I says I want to go into a uh a portal and be able to click to start a script and it moves in a certain pace, so I can follow along and read it. Right? Basic things. And you have that document though, and now you have a guiding post. You create architecture, you see how it's gonna be built. You don't have to know all of it. AI just needs to have it, so it helps it, right? So we conduct the research, we then outline the plan, we have kind of the points. If we're doing the automation, we think about what's the starting point. Is it an email? Is it a message for me? All the jazz. We move on from our outline into, you know, in our scout, right? What's our next thing is you, and this is where we unleash the build. Uh, and this is where we uh go into what we can build as fast as possible. Uh and the reality there is we have to in AI have things testing. You know, I believe that um there's so much variables of AI and its performance and style and function and governance that people have to get their hands on it before we can. No assumptions should be made, right? We can get uh uh, you know, ideally as quickly a minimum lovable, you know, what are they gonna enjoy and enjoy with this um and at least see it and experience it. Get enough cool out of it. Uh and then they go into it, and then they can give all the feedback that you can then leverage right away because that's when we go into testing and optimizing.
unknownRight.
SPEAKER_00That's the T. Uh that's the T. So testing and optimizing, yeah. And we test for function, for style, and for governance, kind of like I mentioned, right? So style, you know, does it say, you know, so more so function. Does it do what it's supposed to? Did it answer it? Right? Um, did it answer it right? Style, did it answer it the way you want it to answer it? Is it the right format? Does it look way, sound like you? And then governance, of course, is making sure that it's happening secure, but also in an efficient way, right? So we move on from that testing environment where we can then come in this feedback loop into our S, which is when we ship it, right? This is where we create SOPs, playbooks, wrap it around. Now, in this instance, you were asking about this example. You know what we ended up building is a full-on uh AI agentic platform that has multiple rooms in it, in this app, right? If you think of a house as multiple rooms, one is where you go and you do a sales role play. It's trained based off all their knowledge, it has an elevating level of difficulty. You know, if you do a cold call with it, it's kind of brutal because it is cold, right? They're like, hi, what? How'd you get my number? Right? It's very like, you know, it can be all kinds. They choose the scenario, they choose the avatar type that fits what they're doing, then they jump into a role play. After that, it gives them feedback. But more than that, after they're done, because we wanted feedback on it, it goes to six AI agents, reviews all the different components, and pulls back into the platform and then submits it all into seeing how it works. Right? So the idea, you know, it comes into this really full circle of a full solution. And this is an app, I mean, gong stuff, stuff like that costs a hundred thousand dollars. With AI, we can move so much faster and so much affordable and so much more customizable. There's nothing that can, you know, you can't compete with something that's it's like a tailored, you know, so you could buy a nice car or you buy one that was custom tailored made for you, right? That's a very different game. And I think we're there.
SPEAKER_02I've never heard this, you know, this concept of uh SaaS is dead, right? And you hear that and you're like, oh, there's the headline or whatever. But after hearing this, I realize, okay, my options would be let me go buy an off-the-shelf gong to use to effectively trade my sales team, or you know what? Let me throw a little money at Aton, let me give him a couple weeks or whatever long. Like, and and what I find amazing about this, and listeners, this is important for you too, because you may be saying, Whoa, this is all way over my head. A ton didn't get a degree in computer science or coding. He wasn't building websites or apps prior to this stuff. These tools landed. Uh, he started just like all of us do with any tool, he's, oh, what does that button do? Oh, let me try that. And through repetition, he's now to the point where he's building something that I mean, that that's pretty robust. I got a brand new salesperson, they don't know anything about my products. How do I train them? Let them get some practice reps, and let them get deep analysis on feedback all within one application. Like I want that. You know, Ata, one of the examples, uh this was we were speaking at um uh in Salt Lake City uh a couple weeks ago. You just kind of playing around on the laptop and you were asking me about, oh, what are you doing? And I was at the time, one of the things we do for our clients is our prospects, when somebody comes in, we'll run uh research on, okay, this is the company, um, and we do, you know, like a big four style analysis of how would this company and this in this industry benefit from say generative AI being deployed in their business. It didn't take us, you know, it was it didn't take us a long time, but we have so many requests for these coming in there was a backlog, and I was like, man, I can't, that's not a good look, right? So I told you what we were doing, and as you sat there on your computer, I think we timed it, it was two hours. It went from, but we were also eating Chinese food in between. You asked me a few questions, I gave you a little information, you typed a little bit, and within two hours, you had created a front end where I could simply enter the company's URL. And let me think. Okay, so it went from there into perplexity to do context research on the company. It passed that from perplexity into Claude and ran it through a prompt that we use to do these, like to instruct the large language model to do the research. It got the result, and then it sent it to gamma.app, which is a uh tool that we're very, very fond of here at Chief AI Officer, but it will take raw text and it'll turn it into a beautiful presentation. And within two hours, my backlog was done, like you built that thing, and it's all I had to do was copy paste URL, copy-paste, URL. And we had thousands of these created within minutes. So that's how easy this stuff is. Um what I'm interested in, because I've I've kind of gotten started with it, I was like, you know, man, I'm not technical. I have no interest in like getting in the weeds on this stuff. But Google AI Studio came out a few weeks ago and tell you and show you what I did. But what what has the learning curve been like? Because you started even just six months ago, it was harder than it is now. But what was that like what got you started and what what's that journey been like to get to where you are now?
SPEAKER_00Well, I mean, I think the best is truly this is just the best. And I'm not and I've and I'll say this because I did spend more than one should. I've done you know AI training that does exceed, you know, at this point, over six figures of programs and certifications and all this. And I'll be honest you have invested in yourself. Yeah, yeah, personally invested in my own little degree since it came out, you know, immediately started what certification can I have. Well, and of course, chief AI officer was I think probably the the second or or you know, uh up there on the early ones I did. You know, I did a few others, but the the I think the the reality fact, you know, all of that was incredibly strategic and really gave me the picture of how to use you know how AI works. I think the implementations were really most interesting to me was seeing non-technical people AI fy their way of thinking. And I think that's where I started was looking at how all these people um you know in the marketing world were all over this, right? All the course creators, all the whatever. But I was curious to see how are they building it? What are the what are they thinking? What's their process of just thinking through it? And that was part of my education was that just paying to see what people are thinking and how they're thinking about taking their IP. But then that was one of my first projects that I started doing for people, and this was the highest demand thing initially, which was I'm gonna take your IP and AI fight for you, right? I'm gonna make uh a digital you that knows and can talk to people and just quantify you. And it was working really great. I mean, that was the first testimonials I was getting, you know, just nonstop on that stuff, and it was really powerful. But the education was literally playing with it and trying to understand almost how does it, you know, how to take the different interactions of it and Make it, you know, first thing was as human as possible. And as a person, I think because I started my journey in that way of like how can I make it as close to the person I'm trying to replicate, it actually makes a really interesting way of thinking about AI and how it works and learning kind of ways to try and, you know, what does it matter in your perspective? An adjective and its effect on the whole system problem, like all these things that you don't think about, versus you know, the role, but also the perspective, but also the lens and also the experience and also what they read, right? And also examples uh, you know, that you can replicate and how do you use, you know, I knew, for instance, that AI, you know, is NLP and natural language processing. I was like, I don't know exactly what that means, but I know that it used it to process it, so why don't I use that by telling AI to do that for transcripts and start to go through a process of what someone who is really good at that would do. Because this was like kind of a fake it with AI till you make it a version of I'm gonna mimic the experts I need to mimic to get to the results I want to get on these specific things. And I think that's where uh that was the real education, I think, of of just going in there and building, you know, now I've built over hundreds of chatbots and things like that, and it's the there's different types of of work, but the the how AI operates, and I think when we talk about vibe coding, we talk about you know applications, it's again who's really good at building applications? Well, software developers are, engineers are really good at architecting it. Well, if we're saying that AI is taking over development, would it be good for me to understand engineering principles of applications so I can have AI do it in a smart way? And that's where I went on this long, you know, six-month thing with IBM and all this stuff in books. But the reality is most people just need to know vocabulary, language. Uh you know, I think it's uh it's this is where it comes. It's like understanding the words. If you know just enough engineering words, you really start to change the conversation uh of with this these tools, I think. Interesting.
SPEAKER_02Because I I'm not at that level yet. But I took it makes total sense. What are some of the like the low-hanging fruit opportunities for somebody who's like, yeah, I'll go get a replic account, I'll get a lovable account, let me let me try this out. What would you suggest would be a good, not too big of a scope, but something where they can actually build something where they're like, wow, this thing works, and they can do it, you know, maybe on their first or second attempt.
SPEAKER_00Well, it's an interesting scene that you mentioned the Google AI studio, because I would imagine there's actually funner, quicker wins you potentially could have there, because I've already done some like unbelievable. I mean, it's unbelievable. It's amazing. I love it. Again, so all these tools you'll run into the same problem, which is why I'm gonna give my advice in reverse on this, but it'll be the same. Uh, I'll get to the point. Um, the advice reverse is all these tools are really great, and it'll take you to a certain point where you will need to take it to a different notch. How to actually download it, deploy it, connect it to things, make it usable for other people, all these other considerations. However, um to get your first win, I would say a focus, and I'm saying this because you want to take it to a point where I don't necessarily want to scale something out. So I would go into an internal tool, something that you want to use for you, right? For you or for a you know, internal tool for your company, something that other people can log into locally that you don't need to like build this whole monster for, right? And take one particular problem that you think, you know, or or idea, right? You know, to make it a win in your company, definitely make it something valuable. But if you're just playing around and you're trying to get a win, try and figure out something fun that you're trying to solve or do, and then say, okay, this one problem and make sure that that one problem isn't too big of a problem, meaning not too many components in it. When I said before that we disaggregated it into all those components, all those components become different features. So what you would want to do is have a problem, but have a feature within that problem that you want to work on. Say, okay, I want to be able to have Replet answer my Slack, you know, send me all my, you know, create a dashboard for all my Slack messages. All it all I wanted to do is to pull the messages and display them in a nice way, all the ones that involve me in one place. It doesn't seem very exciting, but it gets very cool. But suddenly you have your dashboard where it's all there. You're not jumping between 100 channels, it's all in maybe one dash that looks nice and shows what you got. Yeah, I don't know, just an idea. And having the idea today, because I have all these Slack messages in like six different channels or environments, right? So maybe that's a step one. Just aggregate it. And don't worry yet about some of the other elements. Well, what do I do if that once I have it? Because then if you want to, you can work on a next kind of piece of that puzzle. So, okay, well, I now that I see it and I'm able to pull this information. Well, wouldn't it be cool if it just drafted responses for me? Okay, cool. Why not I have that? Okay, well, what does it need to draft responses? And that's the question you need to ask yourself. What does it actually need to do to do what I want it to do? Right? Then it's kind of like this I want it to be able to draft responses. Cool, what does that mean? Well, I likely need to connect some type of language model, I likely need to have some type of logic, ideally has context on who I am and what I'm doing, so it replies well. But you may want to just start by, again, don't try and build the whole thing. Say, I just wanted to draft a response with AI. It could be a bad response initially, right? But work your way there. So there's ways to go about it. You can architect the whole solution if you're an engineer. But what I'd recommend starting out, solve smaller things and see how they work. And while you're doing them, have AI tell you, hey, how does this work?
unknownRight?
SPEAKER_00Or ask you before you build, hey, what do we need to do to make this work? First tell me, don't do anything yet. Just tell me what it does that need to work, and then it will tell you. You say, okay, that sounds interesting. Tell me more about that. And how does that work? And what else would I need? Uh and I can tell you there's no, I don't think there's a better school than what you could do with AI. Because, and this is maybe controversial, but the reality is you are hands-on with your teacher. You're bit you're in the trenches with your teacher, and if you're asking it the right questions, you're gonna get the right answers. If you're asking you know all kinds of questions, you'll get to all kinds of places. So I think the questions do matter, don't get me wrong, but you're in it. You're actually doing the stuff and you can and you're seeing it's work or not work. You're wondering why. After you do the same mistake five times, but you do it really fast with AI, then you now know. So it's kind of like a rapid learning environment for those who are willing to step into it. But just in a small win, I would say, you know, jump in, solve one small part of the thing that you want to solve.
SPEAKER_02Yeah.
SPEAKER_00Right?
SPEAKER_02Yeah, I'll give you an example for me, because like I'm not quite thinking, I think in AI, as we call it at Chief AI officer, but I'm not thinking as the software creator, the software developer, right? But we have a survey that we run, we sponsor a lot of vestige events. And if you don't know what vestige is, it's like a local CEO mastermind where CEOs get together and they they uh help each other with their business. And we were sponsoring the events, and we wanted to take get information from the people that were there to really just kind of like what's the landscape? Where are people at with AI? Like, what are they thinking? Are they doing it safely? Do they are they even doing it at all? So we had created what we thought was pretty slick, a Google form, right? They fill it out, and at the end we could see all the answers. That was cool. But we're an AI company, man. It's like the best we can do is a Google form. I thought, at least for presentation, what can we come up with? So I went into AI Google Studio. You, you know, listeners, you can try any of the tools that are out there to do this. And I basically got the spreadsheet and I attached it to. And guys, if you've never seen the inside of this, it's like going to Chat GPT. It's just a box and you enter whatever you want to enter. So I entered that and I said, hey, I need a an app that will ask these people these questions and give them, you know, a score at the end. And that was about the extent of it. And I hit enter and away it went. And within a couple of minutes, I now had a web-based interface where somebody could, you know, there were the questions, there were the multiple choice answers, they could select it, and then there was a next button. And I was like, you know what? I'd rather not have to hit the next button every time. So I went back into that box and I said, hey, instead of the next button, whenever they make their selection, move them on to the next question. Oh, but keep the back button in case they want to change it. I hit enter and away it went. Tested it again, and like that was fixed. And I went through it and with, oh, I'd like it to do this. Okay, here you go. Oh, it'd be cool if it did this. Okay, here you go. And my experience was exactly what you just described. As I started getting in motion with it, I started like like the matrix kind of unfolded. I was like, oh, I could do this. Oh, and I just built one today. Um, and I'd love to share it with you, uh, because I'm gonna brag a little bit.
SPEAKER_01Yeah.
SPEAKER_02But one of the things that we do a lot with clients is help them identify where do I start with AI, right? Pilot project identification. And ideally, I I want to be able to have them like press some buttons and help get some help figuring that out. So I built an application now. Um it asks them, it has them come up with the idea and um it has them score on six categories, and it will give them at the end, it'll say, hey, this is probably not one you should start with, or hey, this is great. But AI actually enhanced it and it said, oh, and here's why. So now it gives a summary of why it gave it that. It's it's just been incredible. So that's so good.
SPEAKER_00That's so good. Yeah.
SPEAKER_02Yeah. I I'm not, I'm not, I'm no expert, right? But I was able to do some basic things like that. So as as Aton's talking about this, it's coming from the lens of somebody who is extremely capable of building way more than just like a bell or a whistle on your on your business model. It's but I want you to understand you don't have to spend the next six months, you know, deep in the trenches as he have, it's gotten his 10,000 hours in six months. You can start today, right? But I want you to listen to this interview through the lens of where you can take it, like how sophisticated it can get. So, Aitan, I know that you've um you and I had a conversation, I don't know, maybe a few months ago, and in the chief AI officer community, you know, that was one thing we were subject matter experts on business. Um, a lot of them had gone through chief AI officer certification, they were out there using the tools, but they were self-identified as non-technical. But this is something that you told me, hey, like they don't need to be like they need to be able to do this, but they don't need to be technical. And we started you know, we started letting people know, hey, Aton's gonna do some sessions and we're gonna train some people. Why don't you explain to me what like who showed up, like the rookies at boot camp, who showed up? Um what was their kind of experience, and what did you turn them into in X number of weeks or whatever it was?
SPEAKER_00You know, um what a fun thing that was and uh and is. And I think it's a you know what we what we what we did, right? And by the way, it's not that I I wasn't trading on this for six months, guys, just to be fair. It's been a few years now over the last six months, just to make sure that no one's uh you know. Uh but uh with that being said, it's not like before that, like we said I was in marketing and whatnot, you know, and you know, 10 years military. We were using AI, but you weren't doing anything technical. Yeah. No. Oh yeah. Well, it's it became certainly it became more and more technical, like rapidly over the last year, let's say, right? So let's do this. So um what we did, so we launched a program, and this program was to create builders, because really there was a few reasons. One, I couldn't believe that not enough people from our community are building, uh are building solutions yet. Um, there was too many conversations that end with a chat GPT demo. And there was not like how do I actually take this and make this into something. Um, and as someone who was working as a consultant in this and and selling services, I also noticed that it was very hard to sell stuff that remained in a in a in a chat window versus here's a deliverable you can have with AI. Um and it moved me into us into that discussion of like how can we get more builders who can actually help us out and see how it does. And and we said they could didn't need to be technical because, well, I wasn't when I started, certainly, right? It was kind of an idea of principles. Uh, and I think mindset too. I think not you know, I think there's a level of starting from this place of it is a gift not to be too technical right now. Uh and it's a gift. It's a gift. Because when you're too smart, which a lot of us have been in our own environments, we get stuck. But when we're completely dumb in a new environment, we're able to do we think we can do more than we can. And ideally, if we embrace the fact that we just let that we think that, AI actually can do more than we think we can. Now, with the right frameworks, that's what really enables it. And this is really funny, right? Because um, I think some here, I know, if if people are listening, if you've ever seen someone like a person like, man, I'm so much smarter than that person. How the hell are they doing that? Right? This person's dumb, you know. I want to be that, I wanted to be that dumb person for all the developers out there and all these people who are hyper-technical, build these things, and think that someone like me couldn't possibly be doing what I'm doing. Uh, and it was fun. And it is fun, and I think that's the part is like you don't have to be technical. You can actually take your business experience and become the asset that all these companies out there are needing. And you know this as well as I have. So when we started this program, right, I had all sorts of people. There were a couple that were unbelievably technical, come from like an IT, you know, super developers. Yeah, yeah. 90% of the 90% of the people had zero technical um history. A lot of them, I would say, um, you know, I don't know, 90%, a good amount didn't have any technical, right? And then somewhere in the middle, kind of, you know, all of them were were were certified, you know, chief AI officer folk. They knew AI strategically, certainly, uh in and out in and out. Um, but taking it into that like physical realm of building and developing and going in and not being afraid to like, we're gonna build something right now. Um, you know, the the testimonials we've seen and the people who are just, you know what? Um you know, the step-by-step approach that we take that we took in this course and that I took in this course was based off their feedback, you know, I can I can run on sometimes, uh, creating step-by-step playbooks to really take anyone. And I think that was what they were getting out of it, is this kind of implementation ability of taking, I can now build things. And many people built their first like chatbot, so they didn't just build on a chat GPT, they turned into a product, right? They took it, now they can put it in a nice widget on a website, um, and it's a product. They and and even though that might be simple for some of the listeners, for some listeners be like, how the hell did they do that? The idea of productizing AI into I take this custom thing and now I can make it to your brand on your site, on your environment with your materials, really powerful tool already. But giving people all these, you know, they were able to go now, and you know, this is from what they've you know, what I've seen them say. I'm not just you know saying that, but it's a uh they can now build, you know, a lot of them can now go out and build. I'm not gonna say that everyone, you know, whoever was following along and sees all those things and has the playbooks can now go and build. Top of that, we give them, you know, I gave them all these custom GPTs and and and well, and not it's not GPTs, it's using cloud, but it can build you know a lot of this stuff. So they got tools to do it. So I would say the the funnest thing was the confidence of seeing people actually go in and build stuff who have never done anything like that before. Yeah, that was the funnest thing for me by far, I think, was seeing like, all right, I'm just gonna do it. I don't know what I'm doing, but I'm gonna go into it and see what happens. You know? Yeah. Were they surprised that they were able to do this? I would say there were a lot of people who were um, you know, I think some people were surprised. And some people were uh it was more like a I I think, as I mentioned, I think there's a way of like, you know, if you get the right framework, you or learn something new, you're happy. You do have you have you're unblocked in that way. But you still have to get the biggest barrier of of certain things. You have to move, you know, blindly, you know, or in a way for a bit, especially when you don't know what you're looking at. And you have to trust that you're following a certain step. Every entrepreneur, for instance, knows this, right? You're just looking at one step ahead of you and you don't necessarily know where it's landing. You don't know. But if you just if you keep going forward, you'll figure it out. And you do, and everyone does who does that, right? And I think in AI, if when you're non-technical, that's what it feels like. So I think they there was this level of seeing people embrace this uncertainty of sorts, but within it, learn a new trusting and their new steps and skills that they end up somewhere that is really cool, even if some of the steps they did not know what is going on yet, right? And it's kind of like a process of yes, they have all the frameworks and tools now and they know a lot more of what's happening, but in the process, that moment of I'm gonna, I know that I need to do this, I'm gonna tread through and just see what happens, and seeing you know, people build chatbots, agents messing around, you know, someone had a problem in the community if they built a little app and we did a deep dive on it in a community session to like how do we actually all right, let's build this app. And I built it in an hour, right? Or like a version of it that's almost there in an hour. Uh and it was more to unblock, they knew where they were going. You know, it was just how to ask the right question. I didn't do anything technically different. I didn't come in with some code that I, oh, it's just this and that in a code. Nope. I just asked slightly different questions that led to the result that AI knew how to do. The mantra, I think, you know, if someone figured it out elsewhere, people we're not, you know, we're all very smart, but AI's pretty smart. If someone figured it out late night drinking cola somewhere, yeah, I can figure it out if you ask the right questions, okay? That's what I think. It could be people might, you know, afterwards they'll get messages. No, how could you? It's like, yeah, we're not all that smart. I mean, people have figured it out. 50% of coding is is Googling. Yeah.
SPEAKER_02You know, this is um because people can go, like, there's plenty of content being produced on TikTok and Instagram and YouTube and all that kind of stuff, and that's great. But if you're kind of just absorbing something here or there, um you'll gain some knowledge, but I really like this idea of a like a set sequence. Like uh, we start with this and then we moved on to that. It makes a ton of sense on how these people were able to go from, yeah, I'll give it a shot. I don't know what the heck I'm doing, to, oh my gosh, like I could sell this app, or I could, you know, this just saved me hours in my role or my department or whatever the case might be. Very cool. Because I again I explained, like, AI Studio was pretty easy for me to use. I I just started messing around with it and built some stuff. Hell yeah. Yeah, absolutely. How, how has it, since you've been working with these tools with previous releases, how how much easier has it gotten?
SPEAKER_00I would say, I mean, so uh Google Studio, for instance, I've played around with it a bit, but I didn't take it yet to like production or internal or transferring it to another company or you know, I haven't yet gone through logistics of what it means to use that as a product that I serve people with. Yeah. Um I would say that the building, you know, getting to a very quick, like really cool thing, hyper impressive like AI Studio, like something crazy, right? Um, I think Replet is really great that's getting more depth, but you can also, you know, uh it's gotten way easier, right? I mean, Agent 3, which is the newest release on Replit, for example, can go much longer and thinks much wider and deeper. So it could build end-to-end certain things if it stays focused, if you know, if you're keeping an eye on it, you know, and you're making sure that it's kind of like what do you tell me about this maybe if it stays focused? Well, because AI, so this is notorious right now in the you know, in any of these building things. It's gonna build what you want and then like 10 other things that you didn't even know or asked for. Yes, okay, that's happened to me. It's like why why I didn't ask for that. Why I don't want even that button? Why would you add that button? Like that's not like over-delivering, that's just like a random button that you just added there. Like, I don't even know, like it doesn't even have any functionality yet. You just added it there because whatever. Right? Or worse, that's what you see. There's a whole lot that you don't see that's gonna cause a whole lot of problems later. So there's you know, prompting to make sure that it stays on point and focuses, or if there's a problem, there is a level where I think the you know, creative problem solving and problem solving as a whole, like just being like a rational person, you know, if it brings up a problem, it's like that can't be the problem. In fact, it's worked on this, this, and that. Look harder, right? If you're trying to solve something, just like being a rational person with it, and not necessarily taking everything, you know, you don't have to be technical to debate it. You have to be like, that can't be there's just no way that that's the issue.
SPEAKER_02You're telling me debugging the code is as simple as saying, hey, this didn't work, fix it.
SPEAKER_00Well, there's a there's a few things, right? Yeah, I mean I would say that's a good starting point. The problem though is AI is gonna find off too often the one thing or the first thing that it finds as an issue, as an example. So it says, Ah, I found the issue, and it declares it. And I hope there's like at least a million people laughing right now, because it's so true. I found the issue, I know what it is, it's this thing. And then you go and you is there fine, fix it then, right? And then it goes and it fixes that and it still doesn't work. And you're like, well, good, fix it again, it didn't work. It's like, well, I found the issue. Yeah, yeah, every single time. Yeah. So instead, here's like one trick that I hope that helps some people, right? One thing that you know, we spoke about in this conference. I mentioned this to people too, but it was to say um as a whole, look. I'm not, you know, the exact prompt I don't have it off my head, but you it's the idea of look for why this is happening, but don't stop at the first thing you see, find all possible reasons for why this is. Happening, prioritize them based off likelihood of being the issue, right? And then create a plan to address them before doing anything and show that to me first. And now you're looking at it and saying, okay, and you could do one of two things. You could accept it because you still don't know what the hell that is. If you do know, perfect. If you don't, you could do one of two things. You could say, one, if you're not sure, explain that to me, if you really want to understand. Two, you could then take that to a different LLM and say, hey, I'm facing this issue in my app. What do you need to know to help me solve it? You can ask that back to your agent. You can bring that context to the to that uh other LLM, you know, Gemini, ChatGPT, whatever, wherever you want to play. Um, and then you can take that context and say, hey, it's it's found these issues. What do you think it could be? Even if it doesn't know the code, it could be just a smart friend who you're calling where you're like a phone friend who doesn't cost you the credits that Replet might cost you to do that analysis and to work deeper. So that's kind of a way to cost to be cost-effective about some of this, right? So working with other ones to help solve some of those issues. But don't stop at the first problem, find all the potential problems, rank them based off what you think is happening. Um, and and that is helpful for those of us, especially if we don't know, you know, if I can't just open the file and look at the code, which these days is you know definitely much easier. But it's if you're a step more technical, by the way, you can say, hey, which file is this in? You know, which files are relating to this? Just show me those files, and then you can copy the code from those files again to another place. Or if you connect, you know, we can go on and on on the different levels. There are more technical that are way you know smoother for technical, but if you're not technical, you can be just fine, is what I'm trying to say. Like there are more faster ways to do it if you are technical on some of this. But if you're not, you'll be fine too if you're just working with it.
SPEAKER_02A couple of takeaways from that. The way that you describe the problem solving, where it's like, hey, tool, identify what why this is probably not working, find all the places that like it reminds me of you know a year ago when people were telling chain of thought prompting. You know, think think about the solution step by step, right? And people got better results than that. So I'm seeing this translation of how I'm prompting with ChatGPT into actually how I direct in a vibe coding environment. A lot of those prompts that that that way of thinking about prompting could translate in. And then secondly, it's almost like the way you're describing this now, you are you're the director and you've got a team, and you tell this team, hey, I want you to build that. And then when it doesn't work, you can go over here to this other team, like you can take it from Replit and go into Chat GPT and say, Hey, how come this guy's stuff isn't working? Right? Like that's kind of what it sounds like here. And that I mean, like that's what I do on a day-to-day basis, you know, running a business or running a team. Interesting.
SPEAKER_00That's if you were to look at my browser like during a given working day, you know, I'm working, what's on my screen right now? You would see Cloud open, ChatGPT open, Manus open, probably Replay or Windsor, and all of them are probably doing something. Um, and it's not because I can't have multiple chats running in one place, and you know, Manus particularly can run whatever X amount of queries at the same time. It's just that I might be, you know, there's things that I enjoy doing with some others. I don't recommend other people do this. It's not cost, it's not economic. In some ways, it's unbelievably economic. But you know, if you're really just starting out and you don't have any cash for this, don't do that. But the reality is, even if you do $20 a month on each, you're still at like $200 a month for pretty much all the most incredible tools on the planet to ever be created, uh, which is wild, right? Um, to run a business, to run this type of business, right? So, you know, you can go again at the at the peak if you're really wanting to maximize, but $20 a month can get you most usage that you will need on most of these. And if you're uh working on some others, you're probably fine. I mean, so for me, it's really just about certain things that I experience or enjoy. I may leverage something like Manus to build a playbook for me on the task that I need to do later on that one thing. So then I can just go and follow a playbook that's been created to me based off documentation that I fed it, versus trying to figure it out when I get to it, right? That's the world of like, you know, if I had a great assistant, right? And this is something that I learned from a CEO I was in a I was a um, you know, I was previewed and I was mentored by, where he was just saying, you know, the best assistants, obviously, they're a step ahead of you, right? They're preparing stuff for you so that when you get there, it's ready. You're not you're showing up in it in the things that been kind of set up in a way that they're thinking ahead and around of you, right? So how can I have AI do that? Is is what I do often when I'm working on all these different projects, is having it figure out stuff for me for later. Because I don't have to sit there, I just say, hey, figure this out. Then when I get to that later, now I have that set up for me. So I don't have to yeah. Um that's just operations thing, you know, like a little operational AI life.
SPEAKER_02No, I'm learning a lot. And I'll tell you, I'll be honest with you, if I didn't, if Google AI Studio wasn't included in our Google Business account or whatever, so for all of you listening, if you're in Google, you've got access to that tool I mentioned just a second ago. If it wouldn't have been included, I don't know that I would have taken the time to say, let me because to me it just seemed like a a big stretch. There's no way that I could do that, right? But because it was free, I got in there and I was like, oh my gosh, this thing's working. So for those of you that are saying, like, oh, that's uh this interesting conversation, but that's not for me. Look, if you got a Google account, you probably have access to Google AI Studio. Just go to, I'm sure Google would prefer you went to Google rather than Perplexity, but go to Google, type that in, and there's gonna be a link. And once you get in there, listen to exactly what Aton said. Just start, figure out what you want to do, the scope, right? And then start typing. And then hit enter and watch yourself. Like, don't break your arm patting yourself on the back, but watch you, watch you build, you know, that's so good, though.
SPEAKER_00It is so good. It is so good, and it's so surprisingly good. I mean, I've uh I I think this is really I I think you know, on a on a broader context, you know, on the macro of this, I think the when you said SaaS is dead earlier. I don't think it's I think it's far from dead. I do think there's a room for a lot of people to build their own personal tools. I mean, there's no reason why in AI Studio where you can't build, for instance, your own freak, you know, full content machine for yourself, right? I mean, you can do images. I already did one that was like a full-on, you know, upload your photo and create headshots for you, right? Um, there's no reason why you can't do your own any of these tools. If you want to take it to the next level and make it into a product, you could do your own everything. I did a branded icon creator just because I needed it. I was like, I need to do an icon right now. It's completely random, but I can do it in minutes, so why not? Right? And I think with that's with Google Studio, by the way. I did a bunch of these with as like messing around with it to see what it can do. Um and Nano Banana, of course, is one of the best image editing and image uh models out there right now. Then it's like, oh, perfect, right? This is you could just implement these tools into it. So I I do think people should just mess around and have fun with it and and see where it goes. But I also, you know, that's I think in a SaaS dead is dead world, I think not necessarily, because taking that into something public is uh still useful. But most importantly, I think people like with expertise that are building solutions to their problems that other people have and just make it easier for them to do the thing. Yeah. Right. So an example might be, you know, that automation I told you about for um AI builders, right? The feedback kind of condenser, which basically, when we're working on agents and and chatbots, you know, one of the things is you get feedback on it, then you have to go and figure out the system problems. And some of these system problems, I don't know about you guys or whoever's listening, but mine, you know, the one that I just was working on earlier is like 18 pages as long. You can get to a point, that's not where we're gonna say, oh, that's not efficient or not, whatever. It's like, okay, you see the bot then tell me. The the the idea, but the uh the the concept though, making a change can affect a lot of things there. So you have to be very careful and you have to discern where is it happening. So it can take some time to to come in and then refine the agent's capability or his way of thinking across rules and tone and all these other things that it affects. And it's like, you know, those who don't know what a system prompt is for are listening, this is you know, the main problem that affects is like personality and perspective and how it's thinking when you get a when it gets a prompt, right? It's like it's personality knowledge or whatever. So uh the goal there, if we're if we're looking at it then, is how can I condense that time from someone giving feedback about it to getting ideas on how to fix it to implementing it, right? So I built an automation for operational use for builders that basically you know there's a form that they submit the feedback on, it collects the feedback, the it injects with the feedback the system prompt that we're working on right now in its current version, sends it to an AI agent. This sounds like a lot, it's it's really not that much, it's like five or you know, four or five steps, right? You know, and when we learn how to do it, it's very simple. But you build this agent, um, it reviews it, it has a structure. I'm gonna look at it for you know, take information from it, I'm gonna understand what the user wanted, I'm gonna look at the prompt, I'm gonna analyze the section, see why this might be happening, dissecting it as a AI uh you know optimization expert might be right, looking at all the different elements and then suggesting an outline for it, and then just creating either a prompt for it to redesign it in a LLM for you, and then uploading all that into a Google Doc for me to review when I'm ready to review it, when I have time, and then I just work with all the things I have to start to work on it already, versus doing that analysis myself, which might take you know 30 minutes to an hour for me looking at it, not to mention we're all human. We might get sidetracked in it. We're like, oh, that what about this thing? Maybe we should add this thing to that. Yeah. So for sure, it keeps it focused on what do I need to do to optimize it based off feedback, which means I can provide a faster, better service for my clients, which I don't need to wait now for a while to optimize their agent. I get all the tools I need based off my expertise already, but given to an AI to help me do that part. So it's kind of it sets me up like an assistant would, right? Here's your brief. Go on and now spend your 20-30 minutes on this, make it go go back instead. So that's a tool, for instance, and in the broader context that I gave to on our AI builders call this week, I gave it to everyone, just so here's the code, use it. Um it's because I think it's that's where it's at.
SPEAKER_02Yeah. This so I was how I'm how I'm contextualizing this is I was looking at my kids' homework and the teacher had written a note, and it was obviously it was obvious the teacher had read the the homework, right? They didn't just like give it a score. And and I was thinking, man, how labor intensive is that? And what you're just describing is essentially you were the teacher grading their homework, like, hey, here's my my project. That would be reading a second grader's homework, like no big deal. Debugging or evaluating somebody's app, dude, you would have spent so much time if you didn't build this tool. And but can you go buy that tool anywhere? No. Do any of your friends have this tool you can say, hey, can you send me the link? No. You had this very specific, like just for you application, and you said, Oh, boy, it'd be easy if I could do this. Tap, tap, tap, tap, tap, and now you're able to do that. It's incredible. They're able to do it without you, which is even more incredible. So, guys, we're getting to the top of the hour. Um, obviously, there's we just scratched the surface of this. Um, I know by the time this episode airs, the next cycle of the AI builders uh effort that you do taking, I don't want to call them non-technical. The new word I've heard is post-technical. Like just being technical isn't even required, right? So I know that that the next cohort, by the time this episode airs, will um already have started. But just in general, if people want to like witness what you're doing, obviously they can join the chief AI officer community and see, because I know you you're posting a lot of like wins and aha's in there. But how else are uh are people finding out about like your thoughts about this this post-technical environment of coding and creating apps?
SPEAKER_00I think the best place truly is the chief AI officer uh community. Every LinkedIn is a good spot where I do uh when I'm not swamped in in projects, I post more there. But the community is where I post all the time, definitely in the builder's community, uh every now and then on the broader community. Uh but I think there's a uh you know definitely a place that would be the place to go, and then LinkedIn would be a second, you know, uh place to do that, which would be, you know, Aton Pollinger, uh as you know, ETA and Pollinger, I'm sure it'll be in the notes somewhere. Uh hopefully. Uh that would be it for the most part. Uh, because the reality is I'm not like an educator, guys. Um, I did the course because we needed more builders and stuff like that, right? So when we talk about you know showcasing doing, I'm actually working in the trenches on projects. I'm building, developing. Uh this is not a uh, you know, I I'm I'm learning to share more about that information and I want more people to know about it and have fun with it. But I'm coming kind of from like, okay, I'm out of my project thing, let's talk for a bit, let's see what I can share with you guys. Then I'll go back into it, right? And that's kind of the flow.
SPEAKER_02Yeah. So, listeners, um, we've covered a lot of tools on here and all that kind of stuff. Everything is going to be in the show notes. Here's what I'd say: go to the show notes, click a link, sign up for a free account for the tool, and play with it. Click the next link, do that. That's that's how you learn AI. You don't learn it by, I mean, you you make discoveries by listening to podcasts or whatever, watching the video, but how you really learn this stuff is hitting the keyboard, clicking the link, asking the question, like all those things in the AI tools themselves. So, Aton, thanks so much for being here, man. This is um, like you said, it's long overdue. I don't think it'll be the last one with the the speed at which this um you know this technology continues to improve. And I think it's only gonna be a bigger topic as more people listen to things like that and say, I I clicked on the link and oh my gosh, I built something, right? So I think that this is just gonna snowball into like a huge topic in the AI, AI sphere. And um, you know, thank you for taking time out of. I know that we were a little late because you were still wrapping something up with a build. Like, you're not kidding. You're actually still, as soon as this ends, I'm sure you're probably gonna go back to building. So um yeah. So you guys heard it directly from the trenches on this one. So again, thank you so much, Akon. Thanks, listeners. We'll uh catch up next year. Take care. Thanks for tuning in to using AI at work.
SPEAKER_01Don't forget to subscribe for more conversations about how to use AI at work.
SPEAKER_02And a special thank you to our sponsor, Chief AI Officer, for empowering businesses with AI education and training. Visit their website for free AI readiness assessment and AI strategy guide to help you get started using AI at work.
SPEAKER_01That's www.chiefaiofficer.com.
SPEAKER_02Follow us on Twitter at the handle usingAI at work, and visit www.usingai at work.com for free resources to help you harness AI in your role.