Minh Nguyen: He Earns $10K/Month Automating Directories With AI

I had the pleasure of interviewing Minh Nguyen on my podcast, an entrepreneur and developer from the Bay Area who’s been at the forefront of leveraging AI to build and scale businesses.

Our conversation covered a fascinating range of topics, from the practical differences between various Large Language Models (Claude, ChatGPT, and the surprisingly powerful Gemini) to how AI is fundamentally changing web scraping, programmatic SEO, and the accessibility of building software products.

What struck me most about this conversation was Minh’s hands-on approach to AI—he’s not just theorizing about these tools, he’s actively using them to build real businesses like Cash On, a Chrome extension that helps real estate investors identify profitable properties in a fraction of the time.

We dove deep into his AI-powered web scraping system that’s gathering data on 16,000+ pickleball courts (a task that would have taken over half a working year manually), and explored how no-code platforms are empowering non-technical founders to build functional apps in days rather than months.

This article includes the main topics we covered, a link to watch the full podcast conversation, and the complete transcript below for those who prefer to read.

Topics Covered

In this episode, Minh and I explored several game-changing AI applications and frameworks:

• Choosing the Right LLM for the Job – Why Claude excels at coding, ChatGPT’s deep research feature is invaluable for learning new topics (like quantitative trading), and why Gemini’s massive context window and low cost make it an underrated option for 80% of use cases
• AI-Powered Web Scraping & Directory Websites – How Minh built an automated web scraper using Firecrawl to gather detailed data on pickleball courts, turning what would have been 1,300+ hours of manual work into a seven-day background process, and the business model behind programmatic SEO directory sites
• No-Code/Low-Code Revolution – A practical comparison of tools like Lovable, Bolt, Cursor, and Replit, including real examples of non-technical founders building functional products (like Minh’s physical therapist friend who built an 80% complete self-diagnosis tool)
• Building Knowledge Chatbots – A framework for turning existing content (like podcast episodes or YouTube videos) into specialized chatbots that can replicate your decision-making process and provide value at scale

Watch Now

Watch the full conversation below, or click here to watch on YouTube.

Full Transcript

Nik Hulewsky: Minh, I’m very excited. I’m excited to learn some AI today. You are coming to me from the capital of the world of technology, the Bay Area. Here’s my first question: what tools do you use—LLMs—and for what?

Minh Nguyen: I use all the LLMs, and they all kind of excel at different things, so I use them for different reasons. For coding, anything coding-related, I’ll use Claude because it’s the best coding model. For general-purpose things, but also just research, I’ll use ChatGPT, especially deep research. That’s been super helpful. I’m actually trying to learn a bit of quantitative trading right now as a side project with a friend. I know nothing about it, but the deep research has been super helpful with that. And then, actually, I feel like this is the one people sleep on is Gemini. For 80% of use cases, I actually feel like Gemini—for people who code—it’ll get the job done just as good as the other models, but at a tenth of the cost. People, I feel like, they don’t know about Gemini, they don’t know how cheap it is. I was talking to this guy the other day who was doing web scraping, spent $80 with Claude in one day, and then I showed him Gemini. He switched over, did just as good of a job, but cost like five bucks to do what he had done.

Nik Hulewsky: No way! Why do you think people sleep on Gemini? Because almost everybody who I have spoken to has said Gemini is not even on their radar. Two people have said—one person said it was their main thing, and then another person was like, “I just use it for the Google Suite.”

Minh Nguyen: So Gemini is actually good at two things. One is it has a large context window, which means it can consume a lot of information. For example, I think it has a million context window compared to other LLMs which only can do up to 200,000.

Nik Hulewsky: Oh, really?

Minh Nguyen: Yeah, which is—so I guess if you translated it, let’s say ChatGPT can ingest 100 pages, then Gemini can ingest five times that, so 500 pages.

Nik Hulewsky: Well, I just know like Claude’s context window is 200,000, ChatGPT’s is around 120,000. I had no idea that Gemini’s was a million.

Minh Nguyen: I think it actually can even go higher than a million, but I think the cost kind of skyrockets or it goes up. I’ve never needed above a million, so I haven’t tried.

Nik Hulewsky: So you are an entrepreneur, you’re a coder. You’ve built a company called Cashon. Tell me just briefly what that does.

Minh Nguyen: Sure. Cashon is a Chrome extension that hooks into Zillow. Usually what real estate investors have to do is they browse Zillow, they copy and paste the data—like the price, the mortgage, and interest—into a spreadsheet and they run their analysis. It probably takes about five minutes per property. But you can only find maybe one in a hundred properties that would be a good real estate investment. So what my app does is it will grab all that data for you, analyze it for you, and then tell you which of the properties on the map are the best ones. It takes a process that can take usually months and compresses it down into a week or two. Best success story is this guy was looking for real estate investments for like two years, couldn’t find anything. I worked with him, we basically identified a market in two weeks, and then within a month, he put in and got two offers accepted on two different properties.

Nik Hulewsky: That’s amazing. And it’s a Chrome extension?

Minh Nguyen: Yeah, it’s a Chrome extension, and I have a free tier that is pretty generous that people can try out.

Nik Hulewsky: The cool thing I didn’t realize this, but there are some companies that are like eight, nine-figure companies and they’re just Chrome extensions. Which is insane to me. Now I’m curious—the money question: how are you utilizing AI? Utilizing it for work? Personally? Did it build, or at least help build, Cashon? Curious to see your use cases.

Minh Nguyen: I pretty much use AI all the time now. Like I kind of mentioned in the beginning, to learn new topics, to do research, to code. I use Cursor to code. But maybe the thing that I kind of want to show you, which I’m kind of excited about, is I use AI now—I built an AI-powered web scraper. Basically, there’s this concept called directories that’s kind of getting popular right now. The idea is if you consolidate information and put a website up, then people will go to your website and that traffic can actually generate you quite a bit of money. Have you heard of AllTrails?

Nik Hulewsky: Yeah, the app.

Minh Nguyen: Yelp is another example. People most often use it as a directory for restaurants. I built—it’s called pickleballcourtsnearme.com. People search it and then there is search volume—like 200,000 plus per month. With that kind of traffic, you can generate easily in the order of five to ten thousand a month.

Nik Hulewsky: Wait, so are you making money off of pickleballcourtsnearme.com?

Minh Nguyen: Not yet. I’m just starting it up. That’s why I built this web scraper. The idea is I built a web scraper that will help me scour the web for pickleball courts and tell you information like: is it indoor? Is it outdoor? Can it be reserved? Does it have lights so you can play at night? A bunch of other stuff. I was helping a friend who was building dog parks near me, and he was basically saying that it was going to take him 80 hours to gather the data needed to build this website to hopefully make something like two, three, five thousand dollars a month. That’s kind of how I got into it. I was like, “Wait, I can automate this for you, man.” It took me about a day or two, but we got it running. I saved him 80 hours, he launched his website, and we’ll see—SEO is kind of a long game, so we’ll see the traffic that you’re able to acquire. There’s a lot of potential with these directories but also just LLM-based web scraping because they can browse like humans and automate what you’re doing and extracting the information you need.

Nik Hulewsky: Is it similar to programmatic SEO?

Minh Nguyen: Yes, yes. Actually, let me—this will be just easier if I share my screen.

Nik Hulewsky: Yeah, I’d love to see it. I’m like really new to all this stuff. I’m trying to learn as much as I can, but programmatic SEO has been really attractive to me because, like you’re saying, it’s a long game. But you’re also taking advantage of either different geographies or different verticals. So I’ve been thinking about AI use cases around programmatic SEO.

Minh Nguyen: Yeah, so this is my website, pickleballcourtsnearme.com. The programmatic is for every state and for every city—this is California, Irvine—there will be a page and this is the kind of data extracted. This is actually Denver, Colorado. You go to whatever city and then there’s a URL that’s already been created. This is Texas, Dallas. All the data is here. When you go to Google Search and you search “pickleball courts near me,” this will say “pickleball in Dallas, Texas.”

Nik Hulewsky: So how are you using AI to generate the information for these websites?

Minh Nguyen: What I first did is I used this platform called Outscraper. I scraped Google Maps for all the locations that have “pickleball” in the name or in the description. That gives me a list of all the locations and it also usually comes with a URL. Since I code, this is in my database, but usually it’s given to you as a CSV that you can open up in your Excel sheet. The idea is you get the pickleball court or the location—like these sports complexes—and then you get the URL of these pickleball court locations. And then what I do is I use this open-source software called Firecrawl. I send it to this URL and I say, “Okay, I need pickleball courts and I want it to look like this.” So I want the URL, I want the email, the phone number, the address. It goes to the website and then it crawls it until it finds the data needed and then it gives it back to me in the format I need to show on my website.

Nik Hulewsky: So the software you’re using—is that AI-enabled?

Minh Nguyen: Yes. Previously, if you were trying to scrape, you’d have to hardcode everything. For example, you could only go to ebay.com and say, “For this product, this is where you get the email and this is where you get the address.” But now with AI, no matter where it is on the page or even if it’s on a different page in the website, it can decide which one to go to. Usually what happens is it goes to a website and then I get all the links on that website. “Here is the homepage and these are the 100 other links on this website.”

Nik Hulewsky: Again, I’m really stupid and I’m just trying to like process through this. The reason Outscraper is able to get all the information that it does from Google is because Google probably has some standardized way of gathering and displaying information. Because Outscraper had all this information before AI, right? And they’re able to do that because it was predictable—you could hardcode and pull all this stuff. Now what you can do with that website information, using context, is say like “go find me the phone number” or whatever. And you don’t have to tell it where it is, but it knows what a phone number looks like so it’s going to go through and make a decision. How cool!

Minh Nguyen: Exactly. You nailed it. Since businesses usually submit their business and they fill out the information, Outscraper can scrape it from Google Maps because it’s always in the same place in the same format. But what happens if it’s not in the same place in the same format? Then you need a human to go do that. Previously, a job like this would have taken many people many months to scrape. In this database, I have 16,000 pickleball courts. If you say 16,000 and let’s say it takes five minutes per, that’s 80,000 minutes, so that’s 1,300 hours. That’s more than half of a working year. And so with AI, I kind of just let it run in the background. It’ll take me about seven days, but I don’t have to do anything. And actually, what I’ve noticed is it actually does a lot better than a human too. Because with AI, you’re basically getting like a college-level graduate—all you’re doing is pulling data out, so it’s not hard. If you read it and then if it exists on the page, the AI can definitely grab it, whereas a human might miss it or a human might be like, “Ah, good enough. That phone number might work,” and they’re just trying to get to the next thing as opposed to getting it right. Have you messed with Operator?

Nik Hulewsky: I have, I have.

Minh Nguyen: You know how slow Operator is? When it’s going in and executing a task, it’s slower than the human would actually do it. Is it the same with this technology? Is it just slow? Is that why it takes seven days? Or is it just a massive data set?

Nik Hulewsky: It’s just a massive data set. I think with Operator, it will get faster over time. But it’s still slow now because it has to be a generalized use case. My guess is it actually could run faster, but it would cost OpenAI more, so I think they purposely rate-limited it. But since for me, I just need to get the job done as fast as possible, I can let it run for as fast as possible. I’m actually more limited by the computer I run it on. And so I could technically rent a computer in the cloud.

Nik Hulewsky: How much coding does it take to do what you’re doing? Like if I wanted to go—not scrape all the pickleball courts, I wanted to go scrape all of the cupcake businesses in the United States. I know the software now, Firecrawl. Could I just go do this? Or what would it actually take for me to do it?

Minh Nguyen: So Firecrawl handles the scraping for you and gives the website in a format that the LLM can do. And if you were going for like an MVP, let’s say “I just want to grab the data only for cupcakes, don’t need a generalized scraper,” you could probably get that done if you were a software engineer in a day or two. If you didn’t, it kind of depends on how good your prompting is, but I think you could still get it done in a couple days. I think the thing with prompting from a non-engineer’s perspective is just you could probably try everything you could within like three days. And then either you’ll get it done in three days or you’re not going to be able to get it done, like you don’t know how to prompt it in the correct way.

Nik Hulewsky: Yeah, I’m just thinking for a business owner who’s like, “Let’s say I have a PE firm and I’m rolling up cupcake shops, and I want to know who all the cupcake shops are that are owned by mom and pops.” I went and got all the information from Google, but now I want you to find the specific owner—some use case where you’re scraping data—how would I do that? Would I just go to Firecrawl and say this is what I want? Or do I have to build an app essentially to run Firecrawl on?

Minh Nguyen: I think specifically for this use case, there isn’t a product that kind of exists yet that can do end-to-end. I am planning to release this as a paid product, but it will cost a bit of money because, just for reference, for the pickleball data set, that will cost me at least $150 and I’m only halfway done. And then it also cost me seven days of running my own computer.

Nik Hulewsky: That’s nothing.

Minh Nguyen: Yeah, no, it’s nothing compared to like if you had hired a VA for like 20 bucks an hour, because that would take them 55 days, which would be way more money. But yeah, I think for someone just trying to get it done, I would say give it a shot. I think with AI, you definitely have a good chance at getting it done. This is just maybe on a little bit more on the technical side. I think you get pretty far as a business owner without any coding knowledge. No, it’s weird because this last week, I had never done anything even remotely technical. And in the last week, I’ve learned what JSON is—I know what it is, I don’t necessarily know the definition behind it—but I’ve written a Python script in Google Colab, which I was then able to take my files and condense them and I was able to get my transcripts cleaned up. And I was like, “This is freaking amazing!” I don’t know what I did, but I know that AI wrote it and it gave me the output that I wanted.

Minh Nguyen: Have you heard of bolt.new and Lovable?

Nik Hulewsky: Lovable, yes. I’ve heard of Lovable, but bolt.new?

Minh Nguyen: Very similar to Lovable—products that will help you build websites without any coding knowledge. My physical therapist friend, he’s been trying to pitch me on this idea for a while to help like people kind of self-diagnose themselves. I kind of gave him a couple steps to do beforehand before I was like, “Okay, before we can work on this together.” But he was like, “No, I don’t want to do all that.” So he just went on—I was like, “Okay, if you want to do it, try using Lovable.” He used it and he built like a product that, without AI, probably would have taken me like a solid month. With AI, probably like a couple days. But he was able to build like 80% of a working website with just like a couple bugs to help people self-diagnose themselves like for physical therapy pains. And that blew my mind, because he’s had like a bunch of ideas over the years and never been able to pursue them, and now he can.

Nik Hulewsky: Dude, I know a lot of people who are like, “Dude, I have this idea for an app.” And then they tell you the app and it’s like, “Yeah, you’re never going to do that because you would need an engineer or coder or somebody to create an MVP for you.” But you can actually—you can actually do that now.

Minh Nguyen: I know, it’s crazy. Actually, it’s like years beyond an MVP now. Like there’s like small kinks here and there, but it’s definitely working and definitely gets the job done.

Nik Hulewsky: What’s the difference between Bolt and Lovable?

Minh Nguyen: I think they do the same thing. I think they’re just competitors.

Nik Hulewsky: What about Replit? Replit, Bolt, and Lovable?

Minh Nguyen: I haven’t actually tried Replit, I’ve only seen videos of it. I think it falls into a similar category. My guess though is that Replit would be a little bit more powerful.

Nik Hulewsky: What about Cursor?

Minh Nguyen: Cursor is definitely—if I put it on a spectrum for software engineers versus people that don’t know how to write any code: Lovable and Bolt are on the side of people who don’t know how to code. Cursor is definitely on the side for people that do know how to code. And I think Replit’s in the middle. I feel like with Lovable, I kind of had to help my friend debug a bit. It didn’t let me edit anything in the UI. It’s really designed for people who just cannot code and just want the AI to do everything. But with Cursor, on the other hand, you need to know how to set up your environment, you need to know what to tell Cursor, but it’s way more powerful.

Nik Hulewsky: If I wanted to build an app, would you recommend I try Lovable or Bolt?

Minh Nguyen: I would say Lovable because now I actually have experience with it, but from everything I’ve heard, they’re equivalent products.

Nik Hulewsky: I’m going to try it. I’m going to try it next week. I want to try to build something.

Minh Nguyen: Yeah, dude, I think you’re going to be kind of blown away. Another friend had another idea that he’s been kind of trying to pitch me to build—like this Reddit analyzer. And then he just built it without me because he didn’t need me anymore. It’s hard to say no to your friends, but now that they can go do it themselves, it’s awesome.

Nik Hulewsky: Okay, if I were to build some product for myself, are there like some parameters you would give me or some framework you would give me to identify what would be a good use case?

Minh Nguyen: The best projects to start are app ideas where you don’t need to aggregate data from across the internet. Like these calorie tracker apps, for example, where like the user can input the data themselves and you can kind of like do everything within the app—those would do well. But the moment you kind of need to reference other data sources, you’re kind of getting into a part of software engineering that is harder because you need to rely on other services. So anything you can build that’s like completely wholly encapsulated in your app.

Nik Hulewsky: Okay, so I have like 130 podcast episodes, and let’s just say that of the 130 podcast episodes, it distills my philosophy on many things. That would be a good use case for building sort of a closed system app. The “Nik Deal Analyzer” for all intents and purposes. It could be like “I’ll give advice on your business if you’re trying to scale” or “I want to buy this business, what do you think of the multiple?” or “I want to start a business, what do you think of this idea?” Potentially it just gives you feedback. But I guess that’s just a chatbot, right?

Minh Nguyen: That would be a chatbot, but you would be surprised how helpful that is. I feel like in this kind of age where you can build anything, you still need a lot of knowledge to do things well. So I’m working with like an influencer—he has a YouTube channel and he helps with advice on programmatic SEO and trying to decide which websites to build. But he has a lot of knowledge in his head that would take a long time for you to kind of comb through YouTube videos or articles to kind of understand. And so I’m actually building something like a chatbot, but basically, you give it an idea and then it will just ask all the questions that he would ask and assess your idea. He doesn’t have the time to meet with a hundred people, but now this chatbot can. I still think that’s really useful.

Nik Hulewsky: So how would I do it? I want to walk through this because my problem is that the context window, even if it’s a million tokens, is call it 20 podcast conversations, and I’ve got 130 of them.

Minh Nguyen: I would start with—and this is kind of how I’m approaching it—is you probably have a framework to analyzing if a business is going to be a good idea or not. Like: what’s the monetization model? How are you going to do distribution? All these things. And then also you can then assess if that monetization model will actually work out or not. I’d say define a framework if you don’t have one already, but then you take your framework and encode that into a chatbot. Or I could take 10 episodes that I really liked and clean the data—get rid of the timestamps, get rid of the “ums,” “ahs”—get it down to a point where I can fit it into a context window and then just say, “Hey, help me codify my deal analysis.” That’s actually what I did for my friend that I’m trying to build like a chatbot for. I took about five of his YouTube videos, I threw the transcript in ChatGPT. There are these free services like YouTube to transcript. I threw it in and then I was like, “Okay, tell me all the best practices and tips that I need to follow when trying to choose a new SEO website to build.” And then it spat out like 20 to 30 things he said. I then encoded those 20 to 30 things into questions and criteria to analyze, and that’s how I made the chatbot. You could apply that same thinking: give an AI your audio files, say “what are the most important things to analyze and how do you analyze it? What are the criteria to make something good or not good or needs to be improved?” See what it spits out and then codify that into the app.

Nik Hulewsky: This is really interesting. This was amazing. This was my last call of the week and definitely of the 30, this is like this is a top three. Very good. Appreciate it, man.

Minh Nguyen: I’m glad, man.

Nik Hulewsky: All right, hopefully you liked that episode. If you’ve made it this far, you’re either really committed or you’re stuck doing yard work and you can’t actually skip on your phone. So while I have you, will you please help me grow the show? Like and subscribe is the simplest thing. If you’ll go the next step, will you leave me a review? Five-star on Spotify or Apple. What that does is it tells the algorithm that this is a high-value podcast because more people are leaving reviews for it, and it then pushes it out to more people. If you do that for me, I would greatly appreciate it, and I’ll see you next time.