Turning Repetitive Work into Revenue with Automation

0:00I was always working with basically myself and a really small team if I had the luxury of a team. And I think that forces you to get creative. When ChatGPT came on the scene, I realized I could quite literally 10x myself by leveraging all these tools and really learning how to use them. One thing led to another, and I was essentially creating AI employees that I could work with. Welcome to the Talking AI Podcast, where we talk AI with both experts in the field and early adopters. I'm your host, Matt Page, and we're here to demystify AI for you so you can get some value from it. Let's talk some AI.

0:37Everyone's talking about AI, tech companies, startups, digital natives, but there's a whole world out there that's not even talking about it at all. And it's not because they don't care, but it's because no one's shown them how AI can actually fit into their world. And today's episode is about the gap between AI's potential and its adoption, especially in underserved legacy industries and domains where the opportunity is massive, but the exposure is very small. And my guest today, Meg Favis Kuhn, a former CPG marketer who taught herself how to build with AI. She's gone from writing content to writing code with tools like Lovable and Cursor.

1:15And now she's bringing AI to places most people aren't even looking, from training farmers to build micro tools to food entrepreneurs. And she has a real example of how non-technical creators can become AI builders and why the real AI revolution might happen far outside of Silicon Valley. Welcome to the show, Meg. Thanks so much. I'm excited to be here. Thanks for having me. This is going to be fun. I love this topic of, because democratization is one of the biggest things around AI, and we're in it every day. So we just assume everybody knows what's going on. It's really not the case. There's so many people not even thinking about this right now, but I want to rewind a bit because what was one of the coolest things about your journey is you were a marketer in a large CPG company.

2:02And somehow you completely really transformed your whole career into building with AI tools, which encompassed a lot of self-teaching and learning and all of that. But talk me through that journey. What actually triggered you to do this, to start playing around with this stuff? Yeah, that's a great question. I wouldn't call them large CPG companies. They were definitely startups and I think the nature of them being small meant I was always working with basically myself and a really small team if I had the luxury of a team and I think that forces you to get creative so when ChatGPT came on the scene I realized I could quite literally 10x myself by leveraging all these tools and really learning how to use them so I think one thing led to another And I was essentially creating AI employees that I could work with.

2:58This is probably way before people even thought about that as a thing, right? Because I know 2025 is the year of the agent, but I'm sure you were doing this even before the beginning of this year, this idea of an AI employee. Probably not the way they're done today, but yeah, a mini version before they were branded as such, basically. Basically putting all the context of what an employee would do under a GPT and using those to have a full marketing and operations team for whatever company or whatever project that I was working on at the time. And I most recently was with a nonprofit managing a really large grant.

3:42And I started using ChatGPT to manage that 60-page grant document. Oh, such a good use case. It was just such a lifesaver. And I think if I hadn't had AI at the time to use it, it would take me hours longer every single day to do my job. And so just the amount of time it frees you up to do things that probably matter more because you don't have to focus on the admin and the logistic part of things. That's where I really see the value of using AI. People can actually express their creativity. Quick break in the pod. If you're listening to this podcast, chances are you've been thinking about how to actually use AI inside your business.

4:27And that's exactly why we built the AI Opportunity Finder. It's a free tool that helps you uncover high impact, tailored AI use cases based on your business, your goals, your pain points, and your industry. No fluff, no generic use cases, just real ideas that fit your business and the ranked by ROI potential. It takes about three minutes to run and it's like having your own personal AI strategist for free. If you want to try it for free, check out the link in the show notes or go to hatchworks.com backslash AI dash opportunity dash finder. Yeah. And I love that, which you mentioned earlier, you had to be scrappy just by nature of working in a smaller company setting and whatnot, which drove innovation.

5:11I've been in the same boat in a lot of ways. You just have constraints everywhere. And then when this crazy magical thing that was the chat GPT moment and then everything else that came after is, oh wow I can really use this so that that's a really interesting angle there and it's funny too I find that I don't know your thoughts here but I feel like it was marketers and coding were like the two big use cases out of the gate when chat gbt hit those were like the two in my mind that just took off you see a lot of marketers that adopted AI first yeah I agree with that But even from an operations standpoint, the math and stuff wasn't as good at the beginning.

5:52Now it's a little bit better. But to rely as a sounding board when you have data and you want to get analytics from it or run a few simulations on what a marketing campaign could look like if you invested a certain amount of money in this based on your past performance. Those types of early use cases really got me excited about how can I push the boundary with using AI for myself as a solo worker. I want everybody that's like listening right now, if you're zoned out in the car to like think of what Meg just said, because so many people just use AI, when it's AI, name your tool, it doesn't matter.

6:37they use it like Google, right? But what you just talked about is like running a scenario for a marketing campaign and all of these different things. That's how you should be thinking about using AI, which is totally different than information. Give me information about something. I guess like a meta hack is go ask ChatGPT what type of questions you should be asking it for your job and your domain that are creative and nuanced and use ChatGPT to its full. potential. Yeah, I think also just using it as a market research partner, because I often work by myself and didn't have the luxury of talking to other people.

7:21So it became me talking to AI. And of course, I'm not saying everything it's saying is right, but it's at least valuable to have discourse when you're working by yourself. Yeah. Okay. I got a fun or an interesting example that ties to that. So I was building just something on the side with Lovable over the weekend and I was hitting this random bug I just couldn't figure out. So I took the project that was in GitHub, I gave it to ChatGBT, went outside of Lovable, went to ChatGBT, gave it all the context, talked about what was going on. And I had it think about the problem logically and where the bug was happening, give a recommended solution.

8:03And then I fed that back to Lovable and Lovable was like, oh yeah, that's ChatGBT spot on. Like they figured it out. And it's almost like it was this weird moment where I was like, wow, I feel like I just got an extra engineer on the team. And now they're like collaborating in a sense, which back to your point of discourse, you can create that discourse even between different AI tools. totally i love that yes in cursor when using the agent you know toggling between different models because basically you're getting different co-workers working on the same solution and you're able to ideate with different solutions based on their personalities because i really do think these models have personalities yeah it's it's interesting to see what's emerging especially like they had the whole clod four in their testing finding out that it's being deceptive and all kinds of fun stuff like that.

8:56But this is a good segue here because like you're a marketer by trade. And then how did you go from that to building with AI? Was there a moment where you just started playing with this? And then I'm assuming you went through this big kind of learning curve of like learning how code works and you may have had some of that experience in your past. How did you make that jump? Because I feel like a lot of people get intimidated. And they just don't start because they didn't get a computer science degree at some big university. Well, yeah, there is definitely that step in between. I would say early last year when I saw on social media, a lot of people talking about automation and how you can use AI and tools like make.com to start automating manual repetitive processes.

9:46and working with these small businesses, I realized, especially like social media automation, where you want to ideate a whole bunch of posts all at once. I loved that concept. And so I think I started with something simple there where something that I'm going to do again and again throughout the week, how can I build myself in automation? And then I started talking to small businesses that I was consulting with and realizing they had specific needs for those types of tasks. So I started building those out for companies like a juice company, for example. I built an automation in make.com where every time they receive a new order form from one of their vendors, it automatically is read by Google or sorry, it's automatically read the content is taken from that pdf put into chat gpt and then put into excel so that it updates their production totals so that they know how much juice to produce that week and that that's totally on autopilot now so something they used to spend hours doing each week they no longer have to do and it's more reliable than a human and we have the data to support that because there would be human error in transferring those emails to the Excel sheet.

11:16But the way we've built it in make.com, there's no room for error there. I feel like this is a spot where a lot of companies get stuck because they see, to your point, on social media, people doing these awesome, cool things with make and in it in all these other tools. There's this gap between that and actually doing something valuable. What's your approach to identifying like, what do you even automate? What workflow do you go after? Because I feel like a lot of companies skip that step and they want to jump straight to the tool and think that it like magically does some AI magic and it just fixes things.

11:53That's a really good point. I feel like if you don't know what you want to automate yet, you're not ready for that automation because then you're trying to fit a solution that doesn't really need to happen. Yeah, you're not probably ready operationally to realize, oh, this is like a paper cut task for me. I'm burning time and energy on this every single week. Also understanding what's possible with AI. And a lot of people are overeager and they say AI can do a lot more than it can at the current stage. So being really realistic about those solutions and working with someone who knows their stuff so that they don't tell you, oh, this can happen.

12:37but it can't really yeah and i think to one point is the domain expertise and i think this is the area too like we kicked off in the beginning there are so many underserved column industries domains groups people all kinds of things that are underserved with ai i think what's interesting about your background is like you've served some of these folks in like you mentioned like farming and things like that, that wouldn't be the first thing that would come to mind as like somebody leveraging AI. What was that experience like working with some of these industries that you wouldn't think of as being like AI native?

13:18And what was that, that whole process like? Yeah. The reason I even went towards that was because I did have the domain expertise in these industries. and so I felt comfortable coming in and trying to warm them up to the idea of AI at all. Also, no one's really paying attention to these industries, honestly, because it's not that lucrative at this point in time. It makes sense if you're someone solo like myself, but if you are a bigger company and you have a lot of overhead, you probably want to charge a lot more to implement these solutions or build the solutions for them. My approach, I'm able to work with someone one-on-one because the training element is so vital because they'll need to either run that automation moving forward themselves and figure that out or understand fundamentally how it's working.

14:13That way they understand the risks of hallucination or whatever that automation is doing for them. You almost need this foundational training of just how to think about and use AI upfront. Because you mentioned things like hallucinations, models, context, all of these different things, which are new terms and new things you have to understand to then apply it. What does that gap look like? Do you actually go through and do training with clients you're serving? And how do you get to that aha moment, I guess, with them upfront? Yeah, I always start with an intro to AI sort of presentation that I do also as a workshop format for smaller teams.

15:02But definitely with a client who just so that we're on the same page as far as their understanding of AI and the tools that are currently available to us. Because that changes week by week. And so I think it's helpful for them to engage with someone like myself who's staying on top of the latest and greatest in AI. That way they're not ending up with a solution that isn't a little bit stuck in the past, which happens often, I think, in this AI race. yeah and i think for everybody listening to like a lot of people have this perception that they're already behind and it's not even worth trying but to be honest like i feel like the starting line has been redrawn and you're never going to be as close to the starting line probably in your entire life i would argue as you are right now so if you're thinking of what do i have what value do I have to add?

16:04I think the point you mentioned earlier is like, you probably have some type of domain expertise for the people listening, like figure out like, where's your niche? What domain do you understand very well? And then if you cross that with AI, like Meg's talking about right here, that's where you become extremely valuable, either in your own organization and company or helping other companies in a sense as well. That's where the big value is if you can cross those two things. Because AI is very interesting in that. Be curious, you'll take care of Meg. But I think it crosses across every industry, domain, everything.

16:41It's all encompassing in terms of how it can be applied and help almost anything in a sense. Yeah, yeah, I agree. And what you just said about domain expertise, you're able to speak the language of that industry a lot better than just a general consultant. And I think that's so important to be able to talk to someone where you've probably lived their same pain point before too. Yes. So you are pretty good at finding the solution for it because you've experienced that yourself. Yeah. At the end of the day, I think the concept of jobs to be done and finding where the pain is and solving that still exists.

17:27You just have this really cool tool that can solve it in a better way. But like speaking of like underserved groups and things like that, you're doing this really cool initiative called Women Building with AI, which I love. I have two daughters at home and I love like that you're focusing in on this area, trying to like be a leader in the space, inspiring other women to actually start leaning into this. But go deeper into how you're thinking about that. What is the program? And then how do like listeners, like where do they go to actually get involved? I think this is a really neat concept you're starting to do.

18:00Yeah, I have an existing community right now called Idea to App with AI, but that's open to everyone. But I realized I want to hone in on women because where I get my information, even, it's all from men. And that's not a bad thing. It's just that I feel like sometimes women are not encouraged to learn this type of technical expertise, even though I'm trying to target non-technical people and show that I was never technical to begin with. I built a few email campaigns and that was the extent of my technical expertise. But if you really put your head down and learn how to do these things by mostly watching videos, honestly, becoming a student of YouTube university, then you can get really far.

18:58And also you don't have to compromise your identity as a woman. And like I do, I'm just a girl like that also happens to build apps sometimes. And I don't participate whatsoever in the tech bro culture. I live here in the Bay Area, but I actually just don't even know that much about that whole what's going on in Silicon Valley. And I frankly don't care. I just know that no one had to give me permission to go and do this. I just decided I'm curious about doing it. I can build a solution or I can solve a problem. The technology is essentially free and AI allows me to ask endless questions to get it done.

19:43Yeah. Yeah. And back to the like new starting line being drawn. And that's the point there. Like anybody can jump in and do this. I love that you said YouTube university, because I feel like, again, so many people are like, what course do I go take? What do I, what thing do I go do? It doesn't exist yet. Like it literally is being, building built as we speak. I think that's why platforms like YouTube, TikTok, Instagram are such great ones to learn from because the stuff's happening just in time. Like that, go follow some good creators and people in the space and start learning there. And then the last point you mentioned too is use AI, ChatGPT, Cloud, whatever tool you like.

20:23And again, just pick one. They're all frankly about the same at this point. And talk to that to help you learn. That's, it's your buddy in learning. I can tell you how many times I've used AI in that format. And I've seen some of the stuff you do going from not technical to you actually, you know what you're doing from a technical standpoint, like working in the code. And it's cool to see how fast you've done that with AI. It's insane to be honest. I'm in the same boat, right? I'm not super technical. I've been on the product side, like a data analytics background. So I knew some stuff, but I was able to accelerate it so much.

20:59with AI. That's awesome. Were you, how, were you just asking questions or how would you say you did that? So my experience was I had this moment, like the chat GBT moment, like anybody else that just started using it and you're like, whoa, that's crazy. I haven't read emails and copy for you and things like that. And then when I figured out it could write code, that's what blew my mind. And I then they weren't like amazing things they were building, but just the fact that it built something and I'll even take it a step back. I remember doing something with chat GBT where I had it write code and that blew my mind.

21:40And on my local host, I had a website pop up and I did this in a training and everybody was like, oh my God, that is awful. That is so ugly. And in my mind, I'm like, who cares? It's like this crazy moment where it broke code. And I felt like just so many people It just went over their head and now look how fast it's accelerated. And I had an understanding of like I've written SQL before and things like that, but past that I had to learn along the way. So I think as I was building things, I would ask whatever tool I was in lovable, or I'd go over to chat GBT and say, what is this thing doing? So that helps me.

22:18Supabase is a great tool because you can visually see the database schema. I think that's super important to understand just how databases work. And again, just ask ChatGBT. You don't need like, and I'm talking way too long, but you don't need the nuanced like syntax understanding anymore. You just need to conceptually understand how things work and AI can help you do that. I'm curious, what was your kind of path getting to the point where you are? Yeah, I'd probably say pretty much the same. just trying out these tools, seeing how fast they could spin something up when you have an idea. That's invaluable is seeing an idea materialize in real time and like minutes later, that is so powerful.

23:05And as far as learning really what the code means, that comes after a lot of time you spend troubleshooting i think yeah most of this is troubleshooting and you have to have the patience and the stamina to just keep troubleshooting it over and over again because it's not going to be it's not a magic bullet still yeah yeah i'll give you one instance too that i've run into before because a lot of people they'll tell the tool lovable whatever to do something and it's just not doing it, but all the person's looking at is the front end. It's like, why is the button not there? I told you to put a button right here, but the reason it's not happening is probably because the underlying logic in the system is there's something broken there.

23:54Yeah. And there's this one prompt that I've used a lot and I can't remember it verbatim, but essentially you tell the AI, okay, tell me what the intended behavior that we're trying to accomplish is. So it's now thinking through that. okay now go look in the front end code back end code database scheme all that what is actually happening and then it's able to compare the two things i think that's super helpful where you almost put it on ai to figure out in a sense that's awesome i'd love that prompt yeah if you could drop that in the show notes or something like that yeah we'll put that one in the show notes i'll get the full prompt and whatnot i'm curious what's let's talk on ai tools like the coding tools you mentioned we can hit both like the automation tools you mentioned make is do you have a preference of when you use like an automation tool like a make or n8n versus a ai coding tool like a lovable or cursor when do you make that delineation of which do i use or do you use both at times sometimes i use both mostly i'll start in make.com and see what i can prototype for an automation if make.com is sufficient where mostly it has to do with does the automation need a human in the loop at some point and if it does then that usually means it needs a front end so that my end user can interact with it and instead of going into make.com which even though it's relatively simple to look at it can still be very overwhelming and so the beauty of tools like lovable is you're creating that front end interface for your very non-technical client to interact with and understand what's going on.

25:36And you can have troubleshooting logs there too that show up as friendly errors and say, this is what's happening. But yeah, so that's, I think, mostly I decide, I try to do something bare bones and make. And if that works, then I'll stop there. And otherwise I'll engage them in a bigger contract probably to create that front end as well for them. On the AI coding tool side, what's your preference there? Like you mentioned Lovable, Cursor, there's obviously V0, Bolt and 32 other ones, Winsurf that are out there. Do you have a preference? Replens another one. What's your take on the AI coding tools that have all emerged?

26:19Yeah. So I used to be diehard Lovable. exclusively use lovable which i still like it but it burned me because i have i'm really strong on making directories i'm really passionate about it because i think that directory solved part of the distribution problem so if you're creating an app not an internal app for businesses but something that you plan to market or put on the app store discoverable yes you want to be discoverable exactly I think starting with the directory is great because you're building in search intent and traffic to your website. And the play there is to have SEO optimized keywords and organic traffic.

27:03But Lovable is not SEO friendly. And that's because Lovable renders on the client side instead of the server side. They're using Vite and React. lovable is at least. So now I like to use v0.dev to start all a project with the next JS framework and it's a server side rendering. And that has done wonders for sites that I have deployed. And the SEO difference is tangible for sure. Yeah. I got to share this one because I refer to the same issue. There's this guide that somebody put together that basically solves this SEO problem with love a prompt yeah interesting yeah i've been through that same journey i went replet i used that then i think there's like v0 then i went through a cursor phase i've but i've come back to lovable as like my go-to mainly because it's i feel like they're doing the best at the user experience side of things like cursor is cool especially if you're an engineer and whatnot but there's something about the experience that they've honed in on.

28:20I think it is great for non-technical folks. Plus it's getting really capable. And like ever since that moment, they started using Cloud4 when that came out. I just noticed this like instant benefit. And it's back to the point earlier, like when I built that website with ChatGBT, like you just have to know that it's just things are going to keep getting better. I think that's the other piece as well. I totally agree. And yeah, the Superbase integration on Lovable is unmatched, like that user experience. And it makes it so easy for a non-technical person to just get started right away with a real backend.

28:57You could do it in like a couple of clicks. Like I remember banging my head up against the wall, doing Firebase with Cursor. And I spent so much time on the authentication and setting up the backend before I got to any of the even fun stuff. I've totally been there and it's so crazy that was less than a year ago because I only started this in October and I remember trying to mess around with Firebase and all these other tools and now it is just so much easier I'm jealous of anyone getting into it right now because they won't have to go through that excruciating pain of getting all that set up but I think it was good character building yeah yes exactly yeah And I didn't mention, but for more full production apps, I do like to use Cursor.

29:48It's definitely a workhorse. Yeah. Yeah. And a lot of people too, I've seen go from platform to platform. That's the beauty of GitHub because you can take your code and if you want to go to Cursor, you can go there and go deeper and whatnot as well. Yeah. Back to the things you're building for clients and whatnot. What's been the most interesting or off the wall thing you've built or that you can also share? Maybe there's not one you can share with NDA and whatnot, but any interesting domains or industries or things where people are solving stuff with AI? Yeah. So I love the catering example that I started last year.

30:27this is one of my first clients and they just happened to have a solution that I was really eager to build, which is they are a catering company in San Francisco. They do corporate catering and they have 10 different order platforms that provide the orders throughout the week. And they have to wrangle all those orders and validate the pricing of every single order that comes in because sometimes those predatory ordering platforms will misrepresent their pricing which cuts into the bottom line of the small mom-and-pop catering business and i think the bigger companies that are doing this operate off the ascension that they'll miss the pricing discrepancies so i guess long story short i built them a pricing validator so that every single order as it comes in it can match it against the menu pricing and determine is this correct or not if it is it'll send a notification or actually no notification if it's correct but if there's a discrepancy they will get a notification saying this needs to be resolved yeah oh wow that's and that's such a good use case of ai where it can actually look and decide those things which that's beauty of AI it's probabilistic it can reason it can leverage and mimic logic in a sense versus before we had deterministic systems where you had to specify things if this then that and there's just so many more possibilities that open up with how AI works in a sense.

32:12Yeah and actually I was using exclusively AI for this but that was expensive and what I mean is like using chat gpt to read the yeah but now because of what cursor can do it's building those functions out for me saying if this then that so that one you were actually able to solve without ai which is great too that's like a good thing like a lot of people think okay i got to solve everything with ai but that's a use case where you actually started there and then you were able to do it without it is that kind of what exactly yeah because then it becomes cheaper And then also AI is not quite that reliable yet.