I Built a $10K/Month AI Agent

My name is Ivan and uh I built Lancer. We managed to get to $10,000 a month. The third or fourth month after launching. This is Ivonne, a builder from Macedonia who ran into a problem we've all experienced. Boring repetitive tasks. This seemed like a perfect case where an AI agent would outperform a human by like 10 times or even more. So he built an AI agent to automate the boring repetitive tasks. In just a few months, he went from 0 to $10,000 MR. He did it without running ads or building an audience. I think the most interesting part of how we grew Lancer is that we used and we did zero paid ads. So I asked Ivan to come on the channel and break down his entire strategy. And in this video, we'll dive into how to find ideas in plain sight in your daily life, how to shortcut the entire process of building an audience and get customers on day one, and the huge opportunity in building AI agents that automate tedious work on platforms. This one you cannot miss. I'm Pat Walls, and this is Starter Story. All right, real quick. If you're watching this and want some ideas on how to build your own AI agent, well, I put a link in the description right there that breaks it all down. We'll talk a little bit more about that later, but for now, let's jump into the episode. Ivonne, welcome to the channel. Tell me about who you are, what you built, and what's your story. Hi, Pat. My name is Ian, and uh I built Lancer, an AI agent that automates the job discovery, qualifying, and bidding process on Upwork. I think the most interesting part of how we grew Lancer is that we used something I call connectors, and we did zero paid ads. And I'm very excited to share the strategy with you and your audience today. Uh we're going to get all into this connector idea, affiliates, whatever you want to call it. I really, really like how you approach that. But before we're going to talk about the idea and what you built, can you explain what Lancer is, how it makes money, and potentially show off some of your revenue dashboards and show that this thing is legit? Yeah, Lancer. It essentially helps freelancers and agencies turn Upwork into an automated acquisition channel that generates between five and six figures for them each month and also saves them over 10 hours a week of very boring and repetitive work. Essentially, we managed to get to $10,000 a month, third or fourth month after launching. Uh it's a classic subscription model. So, I would say it's a premium pricing tier for a SAS. Uh we're offering now three plans which is the pay as you go which you for $79 you get 30 proposals and then you pay $2 for each extra proposal send then you have a light plan which is $300 for $250 proposals $150 for each extra proposal and then an unlimited plan which was initially the launch offer for $500 a month. Okay, I mean this app is super cool. I love the idea of Lancer especially because I have spent time and pain going through and handling things on Upwork. Before we get into that, we're going to talk all about your app, how you grew it, how you built it. But before we do, I got to understand a little bit more about your background. How do you even find an idea like this? I just made a solution for a problem that I had. So for the past 5 years, uh I ran NB masters, which is a software development agency. Uh we did over seven figures and grew to almost 20 full-time employees. Our main client recruiting channel is Upwork, which turned out to be a very underrated source of quality leads. There are around 200,000 jobs posted on platform each month. Since these jobs are posted across 24 hours in a day and it takes around 10 minutes to qualify a job and then write a personalized proposal, this seemed like a perfect case where an Agent would outperform a human in this test by like 10 times or even more. So we initially built as an internal tool and it crushed for us. Then we did a beta test where I invited several friends which are also agency owners as well. And when they both closed like between them three five figure clients within two weeks of using the tool, I really thought we really had something here and decided to launch it as a standalone product called Lancer. And that's that. There there seems to be a huge opportunity right now in this sort of almost platform automation thing. I've talked to a few founders recently. We have a few videos coming out where we help you automate something on Poshmark, on Pinterest, on Upwork. AI is certainly helping with that. What do you think the opportunity is here? And how could other people watching this find ideas in their daily life like you? Yeah, I think there's a huge opportunity here. AI agents and automating like parts of the workflows that live on top of existing software platforms is very powerful. I would say maybe software niche if I can say it. And I I would say this allows you to get to the customer much easier as well because the users are basically live on the platform. On Upwork, you can you can find your ICP very perfectly because it's all public, you know, like it's a it's a public uh marketplace for talent. And I would say uh every platform which uh has a a huge user base is something that you can automate uh and build something on top of. Okay, I agree. I mean there's this sort of uh buzzword that's going around a lot which is AI agents but this is a great example of an AI agent but actually one that makes money and as you said there are so many platforms and so many use cases to build cool stuff here. So for anyone that watches the starter story channel you're probably noticing that there are a lot of similar ideas like this coming over and over. So take note of that. Let's move over to the build how you actually built this AI agent or this SAS tool. How did you build it? So the initial tool that we used it internally in the agency I built it to and I put it together like over a weekend myself and it was far from a commercial ready product. Then we built an MVP over the next 3 months using the text that we already used in the agency which is basically TypeScript on both the front end and the back end using Nex.js JS and NodeJS and then GCP and fire store for the hosting and then we use LMS in two ways on Lancer. Uh this is for job qualification and then proposal writing. I think what he's showing us right now is awesome. There is a huge opportunity in AI agents right now. But my guess is that you might be wondering where do I even start? That's why I want to tell you about this free guide from Mindre. It breaks down five proven paths to make money with AI agents. From landing your first client to building a full consulting business, this guide will teach you how to spot high-value ideas hiding in plain sight. They even give you a 30-day road map on how to go from zero to your first high-paying client. Ivonne, who we're talking to today, figured all this out on his own, but you don't have to. If you're interested, just head to the first link in the description to grab the guide for free. I can't wait to see what you build. All right, let's get back to the episode. I mean, this is the power of LLMs and AI. You know, 5 years ago, you wouldn't have been able to code up in a solution that helped you qualify jobs and then even write proposals to jobs because there's so many nuances and you know, someone wants you to write a secret word. Well, your AI can probably do that and write a really nice proposal. That is the power of building with AI and a huge opportunity right now. We're not going to go as far into that because what I really wanted to talk to you about and one of the reasons why I wanted to bring you on the channel is again because we always talk about distribution. We could talk about building stuff all day, but how do you get in front of users and you have a very unique approach with this app, which isn't, you know, post on Tik Tok or something like that. You approached growth differently. Tell me how you approached growth in the beginning and then what really worked for taking Lancer all the way to 10K per month. uh to me and this is the most intuitive ways for me to sell at least is to identify first obviously what the ICP is and instead of like going out to them directly using paid ads or cold email at at scale I identify the layer above them basically a profile of user who has uh access to them. So basically I call them connectors and essentially convince them to be affiliates in that way convincing a connector to sell your to sell a product. So in in this case, Lancer becomes a high ticket sales call compared to jumping on hundreds of calls to convince people to try out our $200 a month SAS product. The right connector is someone that first of all has a network with a ton of basically people that you've identified as your ICP. And then they also have a great reputation plus command a ton of trust within your ICP in their network and if they're great at sales that's on top of all of this that's a great bonus. So in Lancer's case uh I identified upper coaches as those connectors depending on the coach they have between five and 10 20 paying customers. So new paying customers each month which pay them between 600 uh and sometimes even more than $1,000 to help them get leads on Upwork. They have great reviews on their own Upwork profiles and so they basically have a great Upwork profile themselves. So they have a ton of uh inbound from Upwork and it's very easy for them to upsell Lancer to their customers as the best way to build on Upwork jobs themselves. Okay, this is actually genius. You get to do what you do best, which is developing a great product. It could take years to build an audience in this space. Why not go straight to the source and find the coaches, the experts, the influencers? We've had a lot of people on the channel that have done very similar stuff. There's something to learn here. And I want to learn a little bit more about how you go about finding these, in this case, coaches, Upwork coaches. How'd you go about finding them, contacting them, convincing them to become your affiliates? Most of our growth came from actually just two upper coaches. For the first one, one of our beta users had actually worked with one of them and he made the recommendation and intro and all I had to do was just demo the product. He was blown away basically at what we've built and he started referring every client since then. The second coach was a bit more difficult. I did call outreach to him over LinkedIn and basically I straight up just offer him $1,000 to jump on a call with me and essentially pay him to be our user. As far as we set up the commissions, this is how we do it. If they completely sign a a client, they onboard them and they set them up on online s uh they get 30% commission for a lifetime and if they just refer them, then that's 20%. Okay. So, this connector strategy, I really like how you did it, how you went to these Upwork coaches, but I think a lot of people watching this might be building somewhere something in some other platform. And what I'd love to understand for you is if you were to start over and build a new SAS, a new AI agent, what would be your playbook for starting over if you had to start over right now? I would I guess the first step is always defining the ICP. ICP basically meaning ideal customer profile or prospect. So basically who are the people that your product or service solves a problem for and actually are willing and ready to pay to use it before you actually start billing customers and you have paying users. You don't really you can't really tell who your ICP is. I would I guess define an ICP for a SAS product is essentially one that onboards very easily. So there's very low friction hopefully self onboards and then like they stay very long time or they don't really churn. uh and in our case this has uh turned out to be an agency user more than a freelancer that are doing like a high volume of proposals. So the next step is going a level above that uh finding a type of user who actually has a network a wide network of of your ICP. And basically this is uh you need to ask ask and answer the question is who are the people that have a network of a lot of your basically ideal customers and but they also demand a ton of respect toward them and uh basically the the your ICP tr has a high level of trust. Again for lancer this is upper coaches. So uh step three would be then uh to write and send a pitch or an offer to the connectors. Go very personalized. Do a lot of research about them. Reference things about them and essentially do even like a long Loom video or reference again like some of their work closing one affiliate you get a lot of value. You wouldn't really be doing this for a random subscriber for for your SAS platform. But for an affiliate you can afford to to do this. Then the fourth step would be actually working out the details. So once you get them in a conversation, you need to offer them. What do they get get in return? And I would say here the standard is I would say it's 20 to to 30% lifetime commission. And then depending on the size of their network, if they're very active on social media, the number of followers, uh maybe like existing agreements they have with competitors you can play around with. So this might might be paying them up front. This is usually again like case by case basis depending on what's the affiliate and what's their actual value and how big is their network. and you need to determine like how much value you'll get out of it back. And then the fifth step is essentially just tracking the affiliate and doing monthly pays. So the good thing here is once you close one of these affiliates or connectors, you get like inbound stripe sales or booked calls. What we do is to then track the affiliates if you have multiple affiliates and then do automate payment payouts is we use this software tool uh called. So, this is affiliate marketing software. It's pretty great and that this makes our life very easy. Uh, when it comes to this, this is exactly what I would do if I were to replicate this strategy starting today. Well, thanks for showing that. I mean, just to see this project here, it's super cool and it just gets me thinking of all the other niche AI agents you could create. I mean, if I was watching this video, if I'm this far into this video right now, I would go to chat GPT and look up platforms similar to Upwork that require a lot of tedious manual work where maybe some sort of AI agent could assist you in uh saving time on this platform. There's so much opportunity here. This is such a cool idea. Let's talk about tech stack. What tools and what stack are you using for this app making $10,000 a month? Well, first of all, for the programming language, uh, Typescript as I mentioned for the agency type the same text that we use for the agency which is TypeScript on both front end and back end which is NexJS and NodeJS even though at this point we're all using uh cursor with OPUS 4.5 and we barely touch any actual programming language and then we use uh open router for the APIs for the LMS for hosting we use a combination of hatner and GCP we use various proxy providers for connecting safely to the Upwork accounts. We use elastic search for basically quering the jobs and all the data we store and then told as an affiliate marketing software. Okay, cool. Well, thanks for sharing that. Uh, last question that we ask everyone who comes on the channel, Yvonne, if you could stand on Young's shoulders before you got started in business, maybe even before you got started with your dev agency, what would be your advice to young Ivonne or to anyone who's watching this who wants to build software and build stuff online? If I could give myself from a young from few years ago, any advice, it would be to actually start building software products right there and then. So I delayed jumping into the what I call arena uh to optimize for income and financial security. Uh I think a lot of people watching Star can res resonate with this. For most it's a high-paying job. Uh in my case it was running a deaf agency which was very highly profitable but with a very hard ceiling and a lack of opportunities to learn basically interesting and high leveraged skills. Uh we actually debated with my co-founder to go all in on AI products even back when OpenAI released their initial API. We decided to focus on growing the agency. And even though it made us a bunch of money, I now think it was probably the wrong approach since the problems you get to solve when building and growing a software product these days are so highly leveraged and applicable to almost any online business. Mastering them would be worth more than any income you can generate in a year or two. And I'm talking about even six figure salaries. Uh so I guess anyone watching this and is still debating where they should start, I'd say go for it. like we are still very early in this like tech revolution which is AI and we have decades in front of us where there will be a ton of opportunities to make a dent with this technology. So true. Thank you for coming on Ivon. I mean it just shows you the power of this is really a side project for you. You have your dev agency that you're running but you got it to $10,000 a month and it just shows all the opportunities to build cool AI agents or just apps that help you automate things in your life. There's so much opportunity right now. So thanks for coming on and sharing and being open and sharing the whole playbook. Thank you Pat. Thank you for having me on. Bye-bye. Thank you to Ivonne for coming on to the channel. This is really powerful episode because it shows the opportunity of building an AI and I know that everyone's talking about AI and the AI bubble and all this stuff, but this is just such a great example of a product that couldn't exist before that is extremely valuable and grew really fast. Automating repetitive work on platforms. This example is Upwork, but imagine there could be apps for Fiverr, for LinkedIn, all these platforms that have tedious work. We just interviewed someone who does it on Photoshop. There is huge opportunity right now to automate boring, repetitive tasks using LLM. This wasn't possible even a couple years ago. It's possible now. And that's where the opportunity is. If you want to do something similar, check out Starter Story Build. I put a link in the description. You can click that. You can get started right now and you'll build an app in a couple days and launch it to the real world and get into the hands of real users. All right, I'll put the link in the description. It's your choice if you want to get started. Otherwise, we'll see you in the next one. Thank you for watching. Peace.