College Student Replaced a $500 Tutor With $5 of AI Credits

There was this life coach who you know for business owners that deal with clients a lot of them also struggle with following up with them and I this workflow for her where you know after meeting there's two messages that are sent one that's the meeting summary and then another one that's a reminder to sign the contract. >> You talked about AI workflows and AI agentic workflows. Can you unpack that for everyone? I feel like even Claude is good for debugging and then if you use cursor they have like an agent that can review your code for you before you commit to GitHub >> and I feel like AI does a lot of work that can not only save you time but also like money. >> AI is it hype or help?

Good day and welcome to the show. I'm your host Pete Cooper and our topic is AI hype or help. Today's guest is Elizabeth TR. Elizabeth is a student at Babson College in Boston. What makes her special is that she's also an entrepreneur and an AI consultant. She has been successfully building a business whilst studying at college using the skills that she learned at college in AI. This is going to be great. Let's get into it. >> Welcome, Elizabeth. How are you today? >> I'm doing great. How are you? I'm doing great too. Thanks for joining us. You're coming in from Boston. Is that right? >> Yeah. Um I go to school at Babson College. So that's in the greater Houston area. >> Yeah.

So I was driving through there the other day. Um and I was thinking of you. Boston's kind of a big town. It's at this top end of uh where we sort of operate because we're a little bit further down from you. But when you say school, it always makes me think of, you know, preschool, but you're actually at college, right? Yeah. >> Okay. >> The entrepreneurial college too. >> Yeah. So, what are you studying there? >> Uh, business administration with a concentration in technology entrepreneurship. >> Oh, excellent. Excellent. Yes. I did an MBA. Uh, and I'm also an engineer. So, you and I do have something in common. And what's the college you said you went to? >> Babson. BAB. >> Okay.

What's that like? Um, so actually that college is like number one in entrepreneurship and I'd say like everybody there is just like building something or like super motivated in their own craft. Like >> there's students that are like building AIB B2B SAS building like beauty businesses. It's just like very diverse in a business sense as well. And like people there really do inspire you to be honest. >> Is that why you chose Babson? Because of that entrepreneurial flare that they're promoting. >> Yeah. I mean, um, right now I'm also part of the generator, which is Baffson's AI lab, and we're launching an incubator to upskill students and AI. >> Oh, wow. >> That already have those businesses.

And then we also have like pitch competitions as well. And I feel like there's a lot of encouragement to build something that's your own. >> Yes. And so whereabouts are you through your college degree? Are you first year, second year, third year? How many years is it? Where are you in that? >> I'm an incoming junior. >> Mhm. That means third year, I think. >> Yeah. >> Okay. And then how many more years do you have? >> Um I have like two more years basically. Like once I complete my junior year, I'm going to be a senior. So >> yeah. >> So it's like a four-year degree. Yeah. >> Okay, cool. So, you're about halfway through a four year degree.

Now, we spoke the other day and what really got me interested and wanted you on the show is that you're doing something outside of that in your own time, I suppose. Um, tell us about that. You've got an entrepreneurial venture you're already working on. What is that? >> Um, right now I do AI consulting for business owners. So, I've like called up some businesses and I usually ask them like what repetitive workflows they have and then if they have one that can be automated with AI, I'll build it for them. >> Right. So, are you a software developer? Did you come from that side and then you come into this or or not? >> Um, no actually.

I mean, I like I have no coding background and I used to be so scared of coding until I learned how to use like cloud code and then some basic like Python syntax and then that like gave me enough like leeway to just build things that can actually sell. It's >> like a superpower, right? >> Yeah, >> it is, isn't it? I always tell people the story of um there's a game called the wheel of fortune and I was in a networking group and he was playing this game and I built it, deployed it and had it running before he finished the game. So that was like five minutes to build the game using these tools.

Enormously powerful and it makes you feel like you got a superpower especially if you're like me that started coding in the days with assembler and as if you don't know what assembler is you actually have to tell it an instruction to move from one register to another you want to add two numbers you have to put it in another register then you have to bring an operation in it was so tedious and every single thing could have a mistake and it only took one little syntax error and the whole thing would crash and there was debugging tools now. It's just amazing and it's fun, right? It's fun doing it because you can see and you can find your way to a solution. You don't get stuck. >> For sure.

I feel like even Claude is good for debugging and then if you use cursor, they have like an agent that can review your code for you before you commit to GitHub. >> Cursive. Cursive is that tool. How do you spell that? >> Just for cursor. Mhm. >> C U R S O R. Yep. >> Yeah. Yeah. >> And what does that do? >> Um it's basically like VS Code. Um and you can also like like open your code file in it and edit code in there and commit to GitHub. But also they have an agent that you can use to basically like write code for you, review, debug, and then connect your API keys. Except I don't think it's that suggested to just copy and paste it into the >> That's right.

Well, thanks for your honesty, Elizabeth. There's there is a lot of copying and pasting. So, you're not really a coder, but I talked a little about assembler. The way code was written in back in the day was you had a blank sheet of paper and then you had to start writing the code out line by line and then you would have to go you would then you would execute the code. In other words, you would compile it, make sure there was no um no errors in that compiling compiling errors they called them. And you usually always was cuz you made a spelling error or something and you have to go and find it and sometimes the tool would help you and sometimes you'd have to find it on your own.

I remember printing out massive piles of paper and going through line by line through all the code trying to find bugs um and trying to fix them. But you had to know the language. And then these languages got bigger and bigger. So C++ is notorious for being massive, right? And lots of libraries and lot of knowledge you have to know. But that's not how you did it, is it Elizabeth? That beautiful that way. >> So, how did you get started then? Let's talk about how you f what was the first project you did.

Um, the first thing I ever learned was NAN because my old internship made me learn it and that kind of got me introduced to Aentic AI workflows and sort of like the mind behind how an AI agent works. So, I think right now like AI agents are one of the most valuable things to kind of have and obviously learn how to make. Um the the first thing I built was actually a newsletter that notified the whole company about a new innovation within Agentic AI because they were selling Agentic AI solutions and kind of wanted to be ahead of the game. >> So I that newsletter through NAN and I like I just kind of like put everybody's email at the end of the workflow so that I could send.

But then before that, I had to like put in what websites it should scrape in order to get that information and then also just like look for like keywords and like different sources of news to see like where it would show up. >> Well, that would be cool. I could see um our company Skillian using something like that. I mean, we I do get a lot of feeds and other things which seem to be doing something similar. Um, but it would be nice to have complete control like that and which is what you have with when you write it yourself. You can say, "Well, I know the ones I'm going to scrape. I know what I'm looking for. Maybe there's a specific niche that I want to really focus on.

" I mean, maybe you could do that for us, Elizabeth. Yes, >> that sounds fun. >> Yeah. Yeah. So, you mentioned a couple of things our audience might not know so well. You talked about AI workflows and AI agentic workflows. Can you unpack that for everyone? >> Yeah. So, sort of how an AI agent works first of all is it's basically like a separate brain like like okay the difference between a chatbot and a AI agent is that if you tell a chatbot to book a meeting for you or schedule a customer, it won't know what to do, right? But an agent is powered by different specific workflows which can be specialized to your business and then it'll basically act on its own.

So it's sort of has its own brain of its own. So it'll take a task and then it'll execute a workflow and then it'll basically do it for you. >> Right? So I always think of of software in the simplest form. It's like an if statement. So if I get an input, I I give it an output and there could be choices in there, right? So first name entered is you could say to the user, do you want to enter your first name or last name? It's like very simple if statements and it's all just built up on logic, right? But with AI, it's it's got the flexibility to respond, hasn't it? So if I don't need to follow its process, I don't need to follow the software's process.

Enter your first name first, second name first. I can just say, I want to give you my name and it will come back and say, well, would you like to put your first name in or your second name in it would take you through that journey. Um, that's where I see the key difference is. It's it's and I think this is one of the things I love about the chatbot is that it takes you from where you are where the where you want to start the conversation and I think this is where you're going with the coding. You you had a point where you started from and then you just started talking to it. You say okay can you the first thing I'm going to say is can you write me some code that'll run on my computer? Yes.

What would you like to be written? What what would you like to be done? And it'll just take you through that whole process. Which is really great because you don't have to have all this pre-nowledge before you start. You just start wherever you're at. >> What do you think? >> For sure. >> I think that like there's such a big learning curve between like [snorts] like like there like there's a learning curve that doesn't exist anymore now that we have AI. And I do feel like more harder skills, especially early on, like when you're a kid, also depend on like uh where you grew up, if your parents t um had had the money to buy you a bunch of tutors.

So I also think AI sort of like lowers the amount of learning that you said would need to be done to get into software engineering because now it's like not are you like only depending on AI for you to learn like like you don't have to depend on it at all like you can sort of think on your own and then just use it as your own tutor basically >> right like >> that's a really good point it's making something which was traditionally out of reach for some kids and students accessible. So, let's talk about the price differences. So, to get a tutor, um, what would you be paying? $65 an hour, something like that. Yeah. And then you'd probably have to give them a few hours a week.

So, you're looking at 120 a week. Let's call it that. And then a month, you're looking at four times a week. Four time? Yeah. Four times a month, you're looking at what? $500 a month for a tutor. For one tutor for one person, right? I'm sure that would work. How much are we spending on AI? >> Like literally $20 a month for the Cloud Pro subscription. And then that also comes with Claw with um like most things you can do like Claude co-work. And then for chatbt, the $20 subscription also includes codeex which gives you a little bit more um API usage compared to Claude. But like say you have both, that's only $40 a month.

Compared to hiring somebody for like 60 a month or hiring a software engineer to build your website for you, it's significantly cheaper. And I feel like AI does a lot of work that can not only save you time, but also like money. >> Yeah, exactly. I think I I just wrote the numbers down for just to give the audience the sort of breakdown. We're talking about the idea of getting a tutor to help your kids or to help yourself get better at something. $500 a month versus I think we tallied up to something like 40. Let's call it $50 a month, right? $50 per month. That's a tenth [clears throat] of the price. That's groundbreaking. That's what we call an order of magnitude in engineering, right?

It's an order of magnitude cheaper. >> I I agree. >> Yeah, that's amazing. Not only that, but you can do it whenever you're ready, not just because the tutor's booked in. Right. >> For sure. Because like um at least for my assignments, like sometimes I really need to ask my professor a question, but AI like already is able to answer it for me. And like sometimes I do my homework at like really like late hours, but >> the AI is always there. Not like it's not all the time that my tutor or my professor will be there because even for like SAT practice questions like I would have to wait until the session for me to go over it with my tutor.

But obviously now with AI like I can like ask it to walk me through different questions. >> Yeah. I mean if you're anything like me, Elizabeth, you have times in the day when you perform better than others. I'm not great when I first get up, but as soon as I have my coffee, my mornings are usually highowered and I feel really energized. Uh, and then in the afternoon, I'm not I usually do different type of work and then I'm usually good in the evening. So, if my tutor was for example coming in midafter afternoon, I'd be at my worst, but that might be the only time I could get the tutor.

Whereas, I want to do things at my best, which would be after my coffee in the morning, midm morning or in the evening. I think that's another huge advantage with AI is not as it cheaper.

We said we got availability and it can fit in with your schedule which some things you don't have control over like I don't have control over how my my bio rhythm my human bio rhythm works just how I my genetics you know that's how it works other things you like you've got a class you can't it clashes with the tutor always when you have to bring multiple people together it's it's difficult especially if it's physical um I'm not saying that that was the case with the tutor that could be done remotely as well but all right so we've got some real advantages with AI, but you know what intrigues me, Elizabeth, is how has you don't know necessarily what happened before when I was at college, but

what's happening now with AI is everybody just getting AI to do their assignments? How does how has the education system changed because of AI? Um, I'd say teachers are a lot more like strict about using it and they're very worried about the extent to which students are thinking on their own or just using AI to answer all their questions. So, I think there's like different systems in place you can put to really make sure that they are learning.

For example, my computer science teacher um he allowed us to use AI because he knows that in the real world we can use it as well, but he had us um give him screenshots of the prompts we would use so that he can make sure that we're actually talking back and forth with it and we're thinking with AI rather than just using it to answer all our questions. So he literally tell us like you need to at least know like what the code should look like and like really converse back and forth and ask your AI to ask you questions because sometimes it doesn't even understand what you want and you need to really guide the output of the model. >> Okay.

So they're actively well in that case actively encouraging the use of AI and focusing on the prompting and I guess the overall structure of what it is you're trying to create. Yeah. So you would have an assignment which would be to build an app for example and you would use claude and chat to your codeex to build it. >> Yeah. So how do they grade such a thing? Because it seems like everyone would just have fabulous results. >> Yeah.

Um, I actually like I know people that didn't even get that good of a grade cuz they just copied and pasted in like what it should look like and then like one prompt into Claude and he really looked for whether you were thinking with it and then also you can tell if like there's at least some effort put when you use AI >> cuz like there's things that are that you can tell are like guided by your own voice like and the assignment had a lot of like like like a menu requirements you had to have criteras you had to meet as well. >> Oh right, right right >> law. Mhm. >> Right.

So by expanding the requirements um maybe today he's giving you much more difficult tasks especially with the requirement side of things and just to explain to the audience when you build code you have to understand what it is you're building and that's called the requirements or the specification right and as you said the menu so when you click on the menu it drills down to all the different things you can go from you know navigate essentially around a page the development of those requirements is is a big part of software development and I don't think that's necessarily changed much it but the building out that piece has changed enormously with AI.

So he's your computer science lecturer is focusing more on the requirements and making sure you're meeting the requirements because the coding part is is I guess just prompting but but pro good prompting is where the thinking has to be done. [snorts] Yeah, that's great. So you're getting students that are doing that well giving a good result. It it meets the requirements and then you're getting others who are I guess doing less work and they don't get such good marks. >> Yeah, for sure. I guess like that sort of measures the effort that's being put in and he really grades it based on that >> because we also learned a lot like like I learned a lot about good prompting through that class.

So when you're writing the code and you're seeing I mean you can generate hundreds if not thousands of lines of code. >> Are you are you able to scroll your way through it and say yeah I know what that does and I know what that does. Or do you just say well I think the structure is about right. I'm getting the results that I want. I don't need to know the syntax. How are you with that? Um, I think it's important to know syntax to also just save like the amount of prompts you use to make sure that like there's no syntax errors and you can sort of fix like small things on your own. And then I also think that like you need to learn about security as well before you use AI.

Make sure that your MV file is under git ignore which you probably know already, >> right? That's a good point. I mean, how do you I mean, we can get into this area as well, but security is always been a problem. There's always been this war back and forward and of course, now you've got really powerful models which can literally are breaking things which were never broken before. I think they found some vulnerabilities in the very core of Linux which were were hidden in there. Um, and Linux is known for its high security and uh, you know, half the internet or more runs with on Linux, but they found these new vulnerabilities. Of course, they're fixing them.

Um, but we're not able to access that model. I don't know whether you you've kept up with the news about Mythos. You've heard of Mythos and we can't access it because it's just too powerful. >> Is that from Claude? >> Yes. Yes. It's one of Anthropic's models. Yeah. [clears throat] Yeah. Okay. So, requirements, security, anything else that comes to mind about how you know how coding has changed, especially in the university environment. [snorts] >> Um, I'd say um like at my school especially since a lot of people are entrepreneurs, it changed like how fast they're able to build and the speed at which like they can really show their investors something.

However, I've also seen a lot of people sort of ship something that doesn't necessarily work like fully. So, I also think that while AI kind of encouraged this culture of like shipping fast and getting things up to speed, um there's a certain like level of at least like like a technical foundation that you should have. Mhm. >> Even as >> So maybe another word for that would be the quality of the the outcome, the output. >> Yeah. >> I think guess there's a temptation. I know I witnessed it. There's a temptation to oh that looks good.

You know, it's doing all this great stuff and it's got a fabulous user interface, but does it actually do the thing that it was supposed to do properly and do it well? Right. And I think this comes back to that requirements really understanding and prompting the requirements well but then at the back end making sure that you really have met the requirements. Um you can produce enormous amount of stuff I'm going to say with AI but there's nothing to say that it's actually any good. It looks good but it's not is it any good? >> All right cool. Okay. So, let's talk a little bit more about um the business you work. Um your side hustle.

I guess you know they encourage you to have a side hustle, right? I mean, the university must really want you to do that. Yes. And so, did you get a lot of advice setting it up? Did the professors really guide you through that and make suggestions or they just said go and do something and off you go? >> No, I didn't get any guidance on it. It was mostly the people I surrounded myself with at school sort of encouraged me to do that. Um, like I said, I'm part of the AI lab. I lead outreach right now. So, we've also had like partnerships with Anthropic, GitHub, and Curser actually.

>> And you know, like obviously we're encouraged to build with those tools, but also their students that are getting clients um and selling things to business owners as well. A lot of them are like building your second brain or building a business with AI in that community. So I think the people you surround yourself with at school have a big deal to do with it. >> So what was what's the partnerships you have with Anthropic and and you mentioned the other ones. What's the substance of that partnership? >> So they basically give like students access to their tools >> um for a buildathon we host. It's it's not necessarily a hackathon.

It's basically like like a version where you build with AI tools instead. And I've seen it like really show students like just like h like what they're capable of like like it blows their minds how much they can build and then like I've seen it too just by like seeing my friends >> seeing what my friends have built and then that encourages me to build as well. >> So are these these buildathons, hackathons, are they a weekend event, daily event? When do they usually occur and how long are they?

>> Um, we're hosting uh a generator weekend where where you have like networking one day and then you can build on Sunday which is this semester and then next semester we're going to host our buildathon because it takes a lot of effort to host those as well. So once a year, but also in Boston like there's a lot of other hackathons that go on and even in New York, especially during tech week, I've seen people win some like crazy things and I just think they're such a valuable experience. >> That's amazing. So how would somebody get involved? Say a business wants to [sighs and gasps] be a part of a hackathon.

Maybe they've got an idea of their own they want someone to build and they want to throw it into the hackathon. How could they do that? Um like like how how could they find it and >> how could they find it and how could they be a part of the hackathon? >> Yeah. Um there's a lot on Luma >> and Participle and then if you just look it up, there's like tons of universities that host these hackathons. >> And then obviously you want to come with like >> maybe you can come with like one friend that's super technical or you can just get yourself randomly like matched with somebody.

But even like like if you don't know anybody there, it's just such a valuable experience to like meet people that are also building. And then like sometimes they have really cool speakers or they partner with really cool companies. I've seen like Nvidia hackathons in New York. >> Um >> yeah, they also had like an AWS and a Claude hackathon where somebody from Anthropic came for New York Tech week. So they're just like super cool and they give a lot of opportunities. And one of my friends won like 3K at a hackathon. So, they're also like me. >> Cool. So, when is Tech Week? >> Um, tech week is the first week of June. >> Okay. >> In New York. >> And then they also have one in Boston and SF.

>> Okay. So, first week of June in New York. We got one in Boston and we got one in um San Francisco. All right. >> In LA. Yeah. Well, there's a lot of universities and colleges throughout the uh throughout the United States. Thousands in fact. Um I would imagine a lot of them are doing hackathons. So if anyone's in a local area, I mean we're down in PA. We're we're close enough to New York. We could do all that. But for us, there's also a really good college here called the Lehi University. Very much known for engineering, software development.

Um you know if anybody wants to be a part of the hackathons there'll be some with uh these local colleges right that's cool so you've got some background here in this area and then you went out on your own to try and find some business so how did you go and do that [clears throat] >> um I think that the resources that you like just readily have available to you like Babson provided me with like two clients that I've had to be honest like there's this initiative called the G1000 where a bunch of small businesses come in and then we can consult them and AI.

Um that's how I got like two clients and then you like so I think utilizing the resources available to you whether at your job or your school are like super valuable and then also cold calling >> business owners to ask them like if they have any repetitive workflows or if they want to automate anything with AI but also like sending emails as well can help like >> you can use a scraper called Appify to get a bunch of emails. >> So, yeah, I think these are all the sort of marketing tools. And is this local businesses that are close to the college? >> Um, sometimes when I'm back, since I'm back home in Connecticut, I reached out to some um near Connecticut and then also like in New York.

Uh I'm I'm talking to somebody that lives in Darien right now. She's a life coach and she wanted to like automate her content. >> Mhm. So relatively local, right? You can you can travel to all of your clients without too much in the car, right? >> Yeah, that kind of makes sense to me. And I think these are small to mediumsiz businesses. So that to me means, you know, five to 50 people, something like that. Is that right? >> Yeah. But also, it's valuable to get like more well-known businesses, especially those that have like big ir ROI on their employees, so that if they have a workflow that's like super repetitive or that has generated a lot of returns.

If you automate that, >> they can then like focus on other things and then you're getting like even more returns at a faster rate. >> Okay. Not quite. I'm sure I understood that. You said a high ROI, a high return on investment. Is that what you're trying to say? >> Yes. So like if each employee within like a certain segment generates a lot of income or like wait no sorry not income but a lot of like sales or money for that business >> and they do like something that's super repetitive then automating that with AI will basically like generate higher or like more frequent returns on that specific. >> Oh okay cool. So, is that something that you did with one of the organizations?

>> Um, I think I've worked with like some smaller businesses, but like one of my friends who does work with bigger clients said that to look for companies that have like high returns on investment on their employees. >> Yeah. So I think the way I would look at that is to say how valuable the time is for that employee and can as you said replace uh some of their time with AI and the best place to look for that is some of the repetitive stuff. All right, that's stuff that's done routinely. Um they may still need to be involved in some way maybe to trigger it, review it, but a lot of the heavy lifting can be done um by AI. [clears throat] That's for sure.

So tell us a little bit about one client engagement that you had that you're proud of. What did you do? [snorts] >> So my most recent one was for this business owner who has six different bank accounts. I mean, if you have like a lot of credit cards, you could also probably understand. >> And he needed to keep track of all of them, but he was but it took him a lot of time to do it manually. And obviously, his time is valuable cuz he runs a business. And then he's also like a dad. So, >> yeah. And I basically automated this workflow that sends him a text message every day at 6: 45 a. M.

With each line uh telling him what which bank account is is which the balance and then the pending charges within the next 24 hours. >> Oh, cool. >> So that he can like basically put money into the into whatever account he needs before it goes negative. And then it also includes like a shortfall warning if any account is going to go negative so that he can know like which account he should put his money towards. >> Yeah. Yeah. No, I can definitely see that. If you got multiple accounts with different banks, you have to go onto each one, log into each one, uh go to the particular account, go to the particular page within that account and then find exactly that piece of information.

Um I would say that's a fairly general problem. Um the people especially that run businesses and have multiple accounts. So you have your business accounts, your private accounts, and if you're a small business, you're often moving money between the two. Some people have multiple businesses, multiple companies as well. So it's quite a handful to keep an eye on. 6: 45 in the morning seems like he likes to wake up to that news. >> Yeah, >> I'm not sure that I want that news. 6: 45 in the morning. Maybe after my first or second coffee. [gasps] Yeah. So that's cool. So how did you go about making it then?

Um, so first I went in a back and forth chat with Claude to ask like each step of the workflow and then I basically copied and pasted in like a long system prompt first into cloud code after opening a project for it >> and then um I had to connect plaid API so I had to go through a bunch of security questionnaires to get access to using plaid. But Plaid is kind of like an interface to the banks, is it? >> Yeah. It keeps coming up one of the tools I use, but I've never been down that path with Plaid. But it would enable you to get a bank feed, I suppose. >> Yeah. And it doesn't cost a lot.

Like the first couple months are like pretty free and then it only costs like a cent like like 10 cents per like not exactly like like low key less than $5 in API credits probably like every month or two >> really. And so you can connect to different banks, different accounts through plaid and then that will give you the the source data and then did you write an app for him or did you just do something on the web or how did you structure that? >> So through Claude I made a dashboard where he could connect his bank account.

So he just like manually connected it through Plaid on that like dashboard and then I made like a cron schedule to basically send him like the lines each day of like the current balances pending charges >> and I connected like GitHub actions cuz GitHub actions is like pretty like like it's free and then um at first I deployed on Railway but railway costs like $5 a month. So then we just switched to GitHub actions for it and it was like not too hard just because I had cloud code. Cool. Uh lots of good stuff in here. I've been jotting them down. I'm not reading. I'm jotting them down because I want to refer back to them. So the dashboard was uh a web app, right?

Was running just on his computer or is it something that you hosted on a server? Um, it was something that I hosted on a server. >> Okay, cool. And then Chrome is automatic Chrome is a kind of automation tool, right? So that's going to fire things off. >> Like it's basically just auler you put in your code. >> Good good word for it. And then this this thing called rail, was it called railway? What does that do? >> Yeah. Um, it's a good site to deploy your website and run any GitHub actions like workflows and stuff in your code. >> Um, it's just costs money after the first month. So, we just switch to GitHub actions for it.

>> So, GitHub actions that sounds like you're getting into agents now. Is that right? >> Yeah, we Yeah. >> Okay. So in GitHub is if people don't know GitHub is a has been used as a repository for code um and allows people to collaborate and also store their own code right but this github actions is something much more than that now they've obviously moved GitHub to a new level and they're actually working in the AI agentic space now >> I guess I think GitHub actions have been there for a file. >> Mhm. >> Yeah. And then I also had to connect Yeah. Yeah. It it's it's it's good. I I like GitHub actions. >> Okay. So, um and then you had to connect it to his text somehow.

You can send a text from GitHub actions. >> Yeah. [clears throat] >> All right. So, what was the sort of cost of him running that once it's all up and running? Because you mentioned cost already. I actually just got my first API bill. Um, it was on my client's card. It was only $7 after like a few months. >> Wow. So, you're looking at like $3 per month. And so, he's saving himself. I always like to tie it back to the about the value. He's saving himself, I don't know, 15 minutes, I guess, something like that, of also just the annoyance of having to do it every day.

I would say if he's good at it, he had it all set up, 15 minutes, he can now look at it in less than a minute and it's costing him $3 a month. >> Yeah, >> sounds like a pretty good return on his time. Do you think other people would want that tool? >> I think maybe. I think people that like cuz he's also a real estate agent so he has like his money moving a lot of different places. I think people that are as busy as him and that have as much bank accounts as him would really appreciate that tool. >> Yeah. Yeah. Cool. And so how long did it take you to build this this tool for him? >> Uh it took me about 25 Wait, no. I split it with a friend. It took me about 15 to 20 hours to build it.

>> And then you had some help as well. >> Yeah, I built it with a friend of mine. >> Okay. So, I guess in round round terms maybe a week. Maybe a week full time because you got to deal with the client and back and forth and bits and pieces are going to fall out in addition to that because there's always a little bit. So, in a week, he's got something that's saving him. And this goes on forever, right? I mean, I can tell he changes things and he's going to come back to he's got a new bank account he wants to add in. He's going to come back and ask you for those little tweaks and adjustments.

I guess one of the key questions that might come up is, do you think that business owner could have done it himself? Could he have built that himself and done it himself? Um, obviously there was like an AI literacy gap, so no, but I think that the need for AI was there because AI is so personalized and there were certain things that he wanted to see like firsthand in the morning in a certain format that like a simple like website that everybody else uses >> wouldn't really tell you. >> Cool. So, yes, I think I agree with you.

Well, I would answer that if it was from my experience, yes, he probably could have done it over time, but without the literacy, it's going to take him a lot longer and it might be quite frustrating and the end result might not be very good. >> And is his time better applied there doing that himself, or is it better applied employing you or another AI engineer um to to build that? I mean, you called yourself an AI consultant, not an AI engineer. Do you think Oh, actually AI strategy consultant. Do you think there's a difference between an AI engineer and an AI consultant? >> Yes. I think an AI engineer is sort of more on the development side.

Like obviously they would know more code than I do and probably like build more of a technical foundation on top of AI. But I think like AI strategy consultant like being a consultant really requires you to listen especially on meetings like when I'm talking to a client I have to make sure that not only do I understand the problem that they're solving but that they understand it too and the value of time that is saved so that I can then sell the workflow to them. >> Oh cool. You've really crossed on something that I think is really worth talking about and that is just that there's two people here. There's you and your client and it's often they don't know what the solution would look like.

All right, you have to paint a picture. So, you have to listen to a whole series of words. Maybe he had a picture for you, but probably words dig digest all that. Understand that, understand the problem and the context of the problem and then package it up and then put it back in a way that they understand that the technology can actually do. Yeah, that's cool. Could AI do that, do you think, though? >> I mean, AI does have the ability to like understand what you're going through. Like honestly, I've been saying I had to talk through some personal problems.

>> I think that like having a human like really understand what you're building, be able to meet with you consistently to edit it, like AI can't do that. >> Yeah. Yeah, I agree. It requires it requires a certain level of knowledge just to get going. I mean, it's one of the things we do often is just try to take people from where they are to a new level. Um because everybody's at a different level. Some people say they never use it. Um others say they're just at the chat GPT level. Others call themselves an expert even though they've only used it for a day and they can put the cat's head on an elephant and show the picture.

Um it's it's very much an undefined thing, but I think everyone can access it. It's just a question of whether it makes sense to do these things yourself or to find someone like you. Elizabeth, do you want to talk about another another customer problem that you solved? Do you have another one that you want to bring into the discussion today? [snorts] >> Um, yes. There was this life coach who, you know, for business owners that deal with clients, a lot of them also struggle with following up with them. And I this workflow for her where you know after meeting there's two messages that are sent.

One that's the meeting summary and then another one that's a reminder to sign the contract cuz you know like a lot of people after the first meeting or two they'll like show interest in it but like they need more time to think about the contract which makes sense >> but like being able to sort of like automate like all the times you have to remind them also saves you time. >> Yeah. Beauty of Marmont. >> Yes. So, how did what was the structure of that? I mean, you went through plaid and you went through that the dashboard and the crone. What did you use for solving that problem?

>> Um, I used NAT initially because at that time I don't think claude code was released yet >> and um she already gets automatic meeting summaries from Zoom because she has Zoom Pro. So I basically just made a system that took that email, it would look for emails from Zoom. So the trigger was each time you had a meeting from Zoom, you would match it to the name of the client and the one that you meet with and then it'll draft an email um sending the meeting summary and then there's a scheduler to wait two days before also following up with a contract. Wow. Cool. Yeah. Imagine doing all that manually.

Especially you got if you're a life coach, you probably got lots of people, lots of stuff that needs to be handled from each of those. I can see you getting very overloaded with a repetitious task. So, that cuts through that. Well, this is some this has been some really great stuff. Elizabeth, is there anything that you want to talk about today? Because we're starting to run out of time and um maybe there's a question for me even. >> Yeah. I mean, as somebody from like sort of the the older generation, sorry, from like from a generation that that didn't grow up with claw code. >> No, I didn't grow up with C code. I didn't even have a computer. I didn't have internet. >> Oh.

Um, would you say that like even though like AI is sort of replacing jobs now, you like there's more opportunity than there was back then for people to sort of build? Like would you say that the pros outweigh the cons or the other way around? >> That's a really good question. I think there's so much opportunity for people like you, Elizabeth, um, who um, are going through college now and have an entrepreneurial spirit. I see an huge emergence for what I'd call soloreneurs. People that are just going to do their own thing, start their own business or work with in partnership with others. It's another sort of direction to go. But just um that was much much more difficult when I was uh younger.

You could have done it in a trade like like as a plumber or an electrician and I did that too. Um but to be in tech was well it really wasn't a thing. I mean electronics. You could build electronics if you but you know computers were just emerging. Um it was very very much harder because you just didn't have access to all of the tools and information and things that you need to build a business. Um even just knowing uh who to call like like back in the day you'd have to go through the yellow pages and think well how do I form a business? Where do I go to form a business? I can't just type it into the internet. I might ask someone.

I might go to the library and and find a book on it, find some source that tells me how to legally form a business. Now I can get that answer in seconds. So all of the things that you want to form a business are just just become easier and AI is just making it easier again too. But the second part of your question is there and I think this is important too. Some jobs are going to go. I mean I I think it's inevitable. Um some jobs are going to be diminished where you you said them yourself. That's where there's repet repetitious jobs that are done on a computer. That's the target for AI.

And who wouldn't want something that is onetenth of the we talked about tutoring right onetenth of the price. Um that's also convenient and and and quality is set to a high standard straight out of the box. I mean, who doesn't want that? People will switch to that. So, it's going to be more difficult um in some areas. And then I think there's those that are just persisting. So, trades people still need to come to the place and fix the sink. You know, AI is not going to do that. They need medical help health people. Or they still need to put their hands on on you to check whether you got a broken arm or whatever.

Like there are some jobs that just persist with and without uh um AI at least in what I'd say the medium term. All all jobs are up up for debate in the longer term because you know we got robots coming but I think that's that's 20 30 years out for some of these things. Did that answer your question? I I guess so, but like more so like like how do you feel about it? >> How do I feel about it? I'm excited about it. I think it's I'm in AI and I didn't have to be an AI. I chose to be an AI. I I like the tools that it's given me. Just as a personal thing, I'm okay talking to people, but as soon as I write stuff down, I'm not very good at writing. It's just never been a thing.

But now I can create the most beautiful essay, document, email, whatever you like. But my ideas are fine. Getting it down on paper was never my strength. Now I don't have that as a barrier. Um I'm not a huge one for coding. I don't need to do that. So it's given me um some I guess some relief from the limitations that I have. And so that's been very rewarding for me. And so I feel good about that. Do I think AI is going to have create some social problems? Definitely. I think there's certain people are going to find it difficult and challenging to adapt and it's not always their own fault. Um but overwhelmingly I think we're heading the right direction. >> Did I answer that time?

Did I answer the question that time, >> Elizabeth? Okay, so we're coming to the end. I have one question for you. Um AI is it hype or help? >> I think it is help but it's surrounded by a lot of hype and that's comes by people just treating it as a all-in-one solution like oh your your business is going to be successful if you automate everything. No, like you still need people and like some things like genuinely should be done manually. But I'd say that if you know where to implement AI, then like that's that's like worth the hype, you know. >> Okay, well said. Very well said. And you and I are very much aligned on that. Well, thanks, Elizabeth.

Stick around after the show because I want to chat to you a little bit. But um bye for now. Thanks. >> Thank you. >> Well, that was great talking to Elizabeth. What a great guest she was and so young and full of enthusiasm for AI and what she's doing. We learned a lot about how she built her business. Well, it's coming back to the traditional way you build a business. Cold calling, emailing, and networking, especially networking because she got those first few leads for those first bits of business from the network that she had through Babson College. Now, what did she actually do? Well, she actually did some automations for people.

One particular example was an entrepreneur that had his own businesses and many different accounts and he needed to see them on a daily basis. She automated that so literally with one text they could see every single day what was going on. That would not only save time but also frustration also make sure someone's acting on their business. I tell you what, she could build me that app as well. But also we touched on something very important and that is the need to have people still in the loop. There is this idea that perhaps AI is going to do everything for us.

Well, she very clearly articulated that when it comes to being a business that's building AI solutions, how important it is that there's two humans there. One to listen to the requirements and create an understanding of the requirements and then build them back and feed them back to the client. Very much a humanto human interaction. Well, if you want to know more about Elizabeth and what she does, please click the links in the description. If you want to know more about Skillian, please do the same. I hope that helps and thanks for watching.