WEBVTT 1 00:00:00.281 --> 00:00:07.480 This episode is the clearest explanation of Firecrawl on the internet and how you can use it to build a real business that makes you real money. 2 00:00:08.080 --> 00:00:10.699 Firecrawl feels like giving your AI eyes. 3 00:00:11.260 --> 00:00:13.680 Right now, AI is smart, but it's blind. 4 00:00:14.020 --> 00:00:17.940 It can't see the internet, it can't go to a website, it can't grab data. 5 00:00:18.380 --> 00:00:19.800 So Firecrawl fixes that. 6 00:00:20.080 --> 00:00:23.860 Once you see it in action, it changes how you think about building products, 7 00:00:24.261 --> 00:00:28.060 how you think about collecting data, and how you think about what's possible with AI. 8 00:00:28.621 --> 00:00:37.520 In this episode, I break down what Firecrawl actually is, how it plays into your AI stack, and walk you through a bunch of startup ideas that you can make money from it. 9 00:00:38.101 --> 00:00:44.460 I use Firecrawl with ideabrowser.com, and I reach out to them to ask them to sponsor this video. 10 00:00:44.900 --> 00:00:50.780 They said yes, so that more people can see this, get the sauce, and build and make money with it. 11 00:00:51.141 --> 00:00:58.860 If Firecrawl has been on your radar and you just want a clear explanation of what it is and how you can use it as a founder, then this episode is for you. 12 00:00:59.280 --> 00:01:01.880 And if you've never heard of it, honestly, that's even better. 13 00:01:02.100 --> 00:01:10.780 Because what I'm about to show you is going to change how you think about what you can build with AI and where the next 12 months of building is going. 14 00:01:11.203 --> 00:01:12.040 Let's get into it. 15 00:01:13.142 --> 00:01:18.831 By the end of the episode, you're going to understand why AI is blind, 16 00:01:24.320 --> 00:01:28.000 why it needs hands and eyes, why Firecrawl is that, 17 00:01:28.841 --> 00:01:31.880 and why the people that understand how to use Firecrawl 18 00:01:32.021 --> 00:01:34.780 are going to be able to create SaaS apps and software 19 00:01:35.244 --> 00:01:37.380 that are super, super valuable to people. 20 00:01:37.520 --> 00:01:40.220 I'm talking the most valuable software products 21 00:01:40.280 --> 00:01:42.579 are going to be using this data scraping tool 22 00:01:43.384 --> 00:01:46.279 at the backbone because it makes their AI 23 00:01:46.502 --> 00:01:47.320 10 times smarter. 24 00:01:47.980 --> 00:01:49.220 But in order to understand this, 25 00:01:49.320 --> 00:01:50.339 we need to take a step back. 26 00:01:50.860 --> 00:01:53.300 The problem is AI is blind. 27 00:01:54.723 --> 00:01:55.720 If you listen to this channel, 28 00:01:55.881 --> 00:01:59.340 you know that the more context you give to a cloud, 29 00:01:59.721 --> 00:02:01.317 the more context you give to a chat GPT, 30 00:02:01.840 --> 00:02:03.560 the better output you're going to get. 31 00:02:03.960 --> 00:02:06.459 So we know that AI models need web data. 32 00:02:06.901 --> 00:02:12.840 It needs top-tier data to actually go and provide really good outputs. 33 00:02:13.902 --> 00:02:14.820 Why does this matter now? 34 00:02:14.920 --> 00:02:19.120 Well, it matters because if you think about the first era of AI, that was the chatbot era. 35 00:02:19.462 --> 00:02:20.696 ChatGPT just came out in 2022. 36 00:02:21.802 --> 00:02:22.620 It answers questions. 37 00:02:22.861 --> 00:02:24.340 It was cool, but pretty limited. 38 00:02:25.061 --> 00:02:26.940 Then we entered the copilot era. 39 00:02:28.140 --> 00:02:29.980 Cursor, GitHub copilot. 40 00:02:30.321 --> 00:02:32.200 It was faster, but you still needed to drive. 41 00:02:32.300 --> 00:02:34.219 It was you, the human being, that was doing it. 42 00:02:34.901 --> 00:02:36.860 We've now entered this AI agent era. 43 00:02:37.260 --> 00:02:38.840 AI is doing the work for you. 44 00:02:39.040 --> 00:02:42.740 Things like cloud code, it browsers, it researches, it builds. 45 00:02:43.323 --> 00:02:44.560 But it still needs the data. 46 00:02:45.200 --> 00:02:47.420 And Firecrawl is how you're going to get that data. 47 00:02:48.181 --> 00:02:51.040 This is often called the computer use era. 48 00:02:51.881 --> 00:02:55.600 We now have AI agents that can see and control computers. 49 00:02:56.401 --> 00:02:57.960 In the past, it was human beings, right? 50 00:02:58.040 --> 00:03:03.820 We bought mouses and keyboards and we had human beings actually going and clicking and doing things, right? 51 00:03:04.703 --> 00:03:07.000 that's going to be the minority, 52 00:03:07.200 --> 00:03:08.560 as weird as it is to say that. 53 00:03:09.121 --> 00:03:11.340 You have tools like Perplexity Computer, 54 00:03:12.100 --> 00:03:14.619 OpenAI had Operator, came out about a year ago. 55 00:03:15.761 --> 00:03:17.300 AI browses the web for you. 56 00:03:17.721 --> 00:03:20.740 GPT 5.4 beats humans at computer tasks. 57 00:03:21.922 --> 00:03:23.759 Cloud has its computer use API, 58 00:03:24.060 --> 00:03:26.900 screenshots and clicks, it's got full desktop control. 59 00:03:27.440 --> 00:03:30.735 Manus was one of the first to do that. 60 00:03:32.204 --> 00:03:34.360 You have browser use, which is an open source. 61 00:03:36.422 --> 00:03:41.340 All these computer uses, all these AI agents that are going and doing things, 62 00:03:41.500 --> 00:03:42.400 well, what do they all need? 63 00:03:42.480 --> 00:03:44.200 Well, they need clean web data. 64 00:03:44.820 --> 00:03:45.839 And that's Firecrawl. 65 00:03:46.341 --> 00:03:50.200 And the reason I got interested in Firecrawl is because I built ideabrowser.com. 66 00:03:51.140 --> 00:03:54.560 And ideabrowser.com is a place where you have trends 67 00:03:54.923 --> 00:03:56.520 and the best startup ideas on the planet. 68 00:03:57.422 --> 00:03:58.459 And I needed the data. 69 00:03:59.386 --> 00:04:00.460 I needed the trend data. 70 00:04:00.761 --> 00:04:05.480 And we built on top of Firecrawl to actually go and get some of that data. 71 00:04:06.140 --> 00:04:09.500 Now we have the number one startup ideas and trends product on the planet. 72 00:04:10.180 --> 00:04:16.660 And it's all because, in largely part, that we're using tools like Firecrawl to actually go and get that data. 73 00:04:17.721 --> 00:04:24.360 What most people don't get about this whole era that we're in is they think that AI is just chatbots that answer questions. 74 00:04:24.781 --> 00:04:27.000 They think web scrapers are illegal and shady. 75 00:04:27.480 --> 00:04:29.080 They think you need to code everything yourself. 76 00:04:29.443 --> 00:04:31.920 They think data is free and easy to get. 77 00:04:32.560 --> 00:04:36.260 And they think that web scraping is a thing for developers. 78 00:04:36.860 --> 00:04:40.840 But what actually is happening is AI agents are doing work autonomously. 79 00:04:41.781 --> 00:04:44.560 Web data is critical AI infrastructure. 80 00:04:45.183 --> 00:04:45.780 Literally critical. 81 00:04:46.361 --> 00:04:49.240 One API call replaces thousands of lines, 82 00:04:49.720 --> 00:04:52.140 and clean structured data is the new oil. 83 00:04:52.261 --> 00:04:54.660 By the end of this episode, I think you're going to agree by that. 84 00:04:55.541 --> 00:05:13.060 So the people that understand how important the clean data is and how important you can use the clean data and wrap it around a brain, an LLM, and wrap that around a piece of software, those are the people that are going to be able to create the most valuable startups in the next 12 months. 85 00:05:13.520 --> 00:05:18.920 And I think that the people that understand that have a 12-month head start, and that's why I wanted to make this episode. 86 00:05:19.760 --> 00:05:23.080 Traditional scraping versus new scrapers like Firecrawl. 87 00:05:23.140 --> 00:05:26.200 Let's just talk about that so we can understand what the difference is. 88 00:05:26.900 --> 00:05:30.060 The old way of scraping was you wrote a custom scraper per site. 89 00:05:30.301 --> 00:05:32.160 You managed proxies and browsers. 90 00:05:33.363 --> 00:05:35.240 You handled anti-bot detection. 91 00:05:35.560 --> 00:05:37.860 You had to parse messy HTML manually. 92 00:05:38.360 --> 00:05:40.120 The scripts would break when site changes. 93 00:05:40.240 --> 00:05:41.220 This happened all the time. 94 00:05:42.202 --> 00:05:43.360 Basically, it was a massive headache. 95 00:05:43.921 --> 00:05:45.400 Now you just do one API call. 96 00:05:45.540 --> 00:05:47.180 You get clean data back in seconds. 97 00:05:47.180 --> 00:05:51.395 It could work on any site, or I think like 99% or 98%. 98 00:05:53.524 --> 00:05:54.753 some high 90% of sites, 99 00:05:56.161 --> 00:05:58.300 and the AI handles layout changes. 100 00:05:59.140 --> 00:06:01.919 So the way I think about my agent stack 101 00:06:02.907 --> 00:06:04.700 is that every builder, if you're listening to this, 102 00:06:04.760 --> 00:06:07.000 you're probably going to need five different layers. 103 00:06:07.640 --> 00:06:09.300 You're going to need an agent harness. 104 00:06:09.900 --> 00:06:11.640 So that's going to be something like a Cloud Code, 105 00:06:12.180 --> 00:06:15.020 Cursor, Codex, or Idea Browser Pro. 106 00:06:16.464 --> 00:06:17.960 You're going to need something that basically 107 00:06:19.004 --> 00:06:22.800 is handling all the different agents in one place. 108 00:06:23.564 --> 00:06:26.140 Then you're going to need something like a search layer. 109 00:06:26.340 --> 00:06:28.720 So something that's going to go and search different things. 110 00:06:28.920 --> 00:06:32.640 Like Perplexity has a good MCP, Exa as well. 111 00:06:33.641 --> 00:06:35.500 Then you're going to need a web data layer. 112 00:06:35.600 --> 00:06:37.520 And that's what we're talking about today in this episode. 113 00:06:38.080 --> 00:06:41.380 So you're going to use Firecrawl for scraping, browsing, and extraction. 114 00:06:42.701 --> 00:06:46.500 Firecrawl, basically the web data layer your agents need to see the internet. 115 00:06:46.620 --> 00:06:56.160 You're going to need to be able to see the internet, to see the data, in order to provide value back in the form of a startup and software. 116 00:06:56.942 --> 00:06:58.120 You're going to need an ops brain. 117 00:06:58.380 --> 00:07:04.260 So I recently did an episode, I encourage you to listen to it if you haven't already, around Obsidian and Cloud Code. 118 00:07:05.941 --> 00:07:14.740 I don't care if you use Notion, I don't care if you use Apple Notes, but you're going to need some brain for storing your meeting notes, storing your context. 119 00:07:15.526 --> 00:07:17.340 and you can use something like Notion or Obsidian. 120 00:07:17.880 --> 00:07:19.460 And then you're going to have to have some outbound 121 00:07:19.541 --> 00:07:22.120 and audience stack as well, 122 00:07:22.260 --> 00:07:24.060 something like an Instantly and Apollo. 123 00:07:24.923 --> 00:07:26.979 And if people are interested, I can spend more time 124 00:07:27.382 --> 00:07:30.440 and do a whole separate episode on some of these tools. 125 00:07:31.821 --> 00:07:33.498 But today we're going to be talking about 126 00:07:34.162 --> 00:07:36.633 the Firecrawl web data layer. 127 00:07:39.600 --> 00:07:40.179 So what is it? 128 00:07:40.422 --> 00:07:41.180 What is Firecrawl? 129 00:07:41.281 --> 00:07:42.780 What is the clearest way to understand it? 130 00:07:43.222 --> 00:07:46.458 You put in a website, goes through the Firecrawl API, 131 00:07:47.400 --> 00:07:51.960 and you get back a clean markdown, a structured JSON, some screenshots. 132 00:07:52.380 --> 00:07:54.335 And you can feed that to any AI model. 133 00:07:55.460 --> 00:07:55.683 That's it. 134 00:07:55.724 --> 00:07:56.699 It's as simple as that. 135 00:07:57.561 --> 00:07:58.895 We don't need to overthink about it. 136 00:07:59.117 --> 00:07:59.420 Think it. 137 00:07:59.961 --> 00:08:03.300 The way I think about it is Firecrawl has six superpowers. 138 00:08:03.760 --> 00:08:07.900 You can scrape, so you can go and scrape one page to a clean markdown. 139 00:08:08.501 --> 00:08:17.360 So something like, scrape one blog post from gregisenberg.com slash blog post. 140 00:08:18.580 --> 00:08:21.220 You can crawl an entire site automatically. 141 00:08:22.220 --> 00:08:23.120 So what do I mean by that? 142 00:08:23.300 --> 00:08:31.840 I mean, you can go and give it cnn.com and it's going to go and crawl all of the different articles on cnn.com and you'll get that data back. 143 00:08:32.741 --> 00:08:35.880 You can map all URLs on a domain instantly. 144 00:08:36.205 --> 00:08:37.080 So that's super helpful. 145 00:08:37.080 --> 00:08:41.700 There's so much metadata and context into mapping and URLs. 146 00:08:43.224 --> 00:08:45.200 Think about a URL, maybe there's a date in it, 147 00:08:45.280 --> 00:08:46.239 there's a title in it. 148 00:08:46.780 --> 00:08:50.900 Having that map is going to be helpful in some capacity to you 149 00:08:51.021 --> 00:08:52.220 depending on what you're trying to do. 150 00:08:52.901 --> 00:08:54.800 You can go and search, you can use Google 151 00:08:55.021 --> 00:08:58.080 and you can put the full content in one call. 152 00:08:58.582 --> 00:08:59.600 Super, super valuable. 153 00:09:00.621 --> 00:09:03.200 It has an agent that you can describe data 154 00:09:03.362 --> 00:09:04.379 and it goes and finds it. 155 00:09:05.268 --> 00:09:06.216 tell it I want the 50 156 00:09:08.087 --> 00:09:10.740 highest rated Cuban restaurants 157 00:09:11.062 --> 00:09:13.219 in South Florida and it's going to give it back to you 158 00:09:14.025 --> 00:09:15.880 going to give you the most clear data on it as well 159 00:09:16.423 --> 00:09:19.220 and then it's got a browser so AI controls a real browser 160 00:09:20.492 --> 00:09:21.400 super super helpful 161 00:09:21.903 --> 00:09:24.339 and it's three lines of code you can screenshot this 162 00:09:24.561 --> 00:09:27.620 or I'll put it in the description for how to sign up 163 00:09:28.223 --> 00:09:30.860 but basically it gives you a clean markdown of the entire website 164 00:09:31.021 --> 00:09:33.800 for any AI model in three lines of code 165 00:09:35.190 --> 00:09:36.519 this is what excites me about it 166 00:09:37.082 --> 00:09:42.120 so I believe that this is the AWS moment for web data 167 00:09:42.465 --> 00:09:43.319 what do I mean by that? 168 00:09:43.941 --> 00:09:46.100 in 2006 if you wanted to build a web app 169 00:09:46.243 --> 00:09:47.039 what did you have to do? 170 00:09:47.363 --> 00:09:49.300 well you had to go out and buy servers 171 00:09:49.441 --> 00:09:51.140 spend thousands of dollars buying servers 172 00:09:51.463 --> 00:09:53.460 you had to go and manage racks and cables 173 00:09:53.864 --> 00:09:55.339 things would break all the time 174 00:09:56.875 --> 00:09:57.339 all the time 175 00:09:57.883 --> 00:09:59.069 and then one day AWS said 176 00:10:01.764 --> 00:10:05.240 one API call and you can use our servers in the cloud. 177 00:10:06.180 --> 00:10:09.820 Now in 2026, if you want AI to use web data, 178 00:10:10.120 --> 00:10:10.852 what do you have to do? 179 00:10:11.500 --> 00:10:15.240 You have to build scrapers, manage proxies, 180 00:10:15.380 --> 00:10:16.940 manage browsers, deal with security. 181 00:10:17.660 --> 00:10:21.960 Firecrawl says one API call and we got you. 182 00:10:22.743 --> 00:10:24.600 This is a big deal because the companies 183 00:10:24.600 --> 00:10:25.626 that were built on top of AWS, 184 00:10:28.763 --> 00:10:30.280 some of them became trillion dollar companies, 185 00:10:30.381 --> 00:10:31.740 Some of them became billion-dollar companies 186 00:10:31.861 --> 00:10:33.400 and a lot became million-dollar companies. 187 00:10:34.560 --> 00:10:36.399 Of course a lot failed, but the point is 188 00:10:37.746 --> 00:10:40.440 people didn't have to deal with the headaches of servers 189 00:10:40.984 --> 00:10:43.320 so they got to focus on building an incredible product 190 00:10:43.884 --> 00:10:45.960 and those products were able to scale. 191 00:10:46.521 --> 00:10:49.520 Some of the biggest companies of the last 10 years 192 00:10:49.883 --> 00:10:51.517 came because of AWS. 193 00:10:52.740 --> 00:10:55.460 So what gets built on the web data layer? 194 00:10:55.663 --> 00:10:56.760 I'm going to give you some ideas, 195 00:10:58.430 --> 00:10:59.560 not billion-dollar ideas, 196 00:10:59.702 --> 00:11:01.400 but some multi-million dollar, 197 00:11:01.700 --> 00:11:02.831 $1 to $10 to $25 million a year, 198 00:11:03.982 --> 00:11:05.180 $50 million a year businesses 199 00:11:06.267 --> 00:11:08.340 that you can start by understanding 200 00:11:08.986 --> 00:11:10.440 what the web data layer is. 201 00:11:10.540 --> 00:11:12.500 And I think a lot of people are sleeping 202 00:11:12.742 --> 00:11:14.960 on how big of a movement this is. 203 00:11:15.881 --> 00:11:18.840 So let's go into how it works. 204 00:11:20.140 --> 00:11:22.433 So here you are, right? 205 00:11:23.641 --> 00:11:24.199 You're the builder. 206 00:11:25.042 --> 00:11:26.320 You've got this AI agent. 207 00:11:27.061 --> 00:11:29.300 And the AI agent is going to go talk to your brain. 208 00:11:29.481 --> 00:11:32.480 So you can use GPT, you can use Clode, you can use Gemini. 209 00:11:33.501 --> 00:11:36.560 You've got a nervous system, that's the way at least I think about it, 210 00:11:36.620 --> 00:11:38.119 which is your MCP protocol. 211 00:11:38.941 --> 00:11:40.739 And now you have your eyes and hands. 212 00:11:41.081 --> 00:11:42.420 Your eyes and hands is Firecrawl. 213 00:11:42.940 --> 00:11:47.620 Now Firecrawl can go out to the internet and it's going to get back clean data 214 00:11:47.680 --> 00:11:53.600 and you're going to use that data to wrap it around products and services you sell. 215 00:11:53.801 --> 00:11:54.979 So this is the big idea, right? 216 00:11:55.060 --> 00:11:57.100 You've got brain, you've got nervous system, 217 00:11:57.521 --> 00:11:59.459 and you've now got eyes and hands. 218 00:12:00.941 --> 00:12:03.800 Of course you can go and do it yourself, scraping. 219 00:12:04.020 --> 00:12:06.240 You can use Playwright or Selenium. 220 00:12:07.444 --> 00:12:09.320 The bottom line is, it's just going to be a lot of work. 221 00:12:09.540 --> 00:12:11.540 I'm trying to do the simplest thing possible. 222 00:12:11.781 --> 00:12:15.980 The reason I like Firecrawl is it's one API call, 223 00:12:16.160 --> 00:12:18.260 proxies are built in, anti-bop built in, 224 00:12:18.400 --> 00:12:19.899 the AI extracts the data for you. 225 00:12:20.341 --> 00:12:22.520 It's just less headaches than actually going in 226 00:12:22.560 --> 00:12:23.380 and doing it yourself. 227 00:12:24.086 --> 00:12:25.820 and you've got the browser sandbox 228 00:12:26.065 --> 00:12:26.739 which is really cool 229 00:12:27.851 --> 00:12:28.720 so the browser sandbox 230 00:12:29.043 --> 00:12:30.333 it's a secure way to have 231 00:12:32.270 --> 00:12:33.320 Firecrawl fill out forms 232 00:12:33.462 --> 00:12:34.400 click buttons and links 233 00:12:34.562 --> 00:12:35.740 handle logins and auth 234 00:12:35.862 --> 00:12:36.879 navigate pagination 235 00:12:37.323 --> 00:12:39.460 you can watch live as your AI browses 236 00:12:40.066 --> 00:12:41.460 stay logged in across sessions 237 00:12:42.067 --> 00:12:43.080 it's really crazy right 238 00:12:43.665 --> 00:12:45.299 so think about it 239 00:12:45.624 --> 00:12:46.579 in a world where you can go 240 00:12:46.620 --> 00:12:48.540 and you have these hands and eyes 241 00:12:48.683 --> 00:12:49.480 out there on the internet 242 00:12:50.508 --> 00:12:52.060 what are the big ideas that you can build 243 00:12:52.161 --> 00:12:53.612 and we're going to be talking about that soon. 244 00:12:58.107 --> 00:12:59.839 The way the agent endpoint works 245 00:13:00.305 --> 00:13:01.340 is you type in a prompt, 246 00:13:01.580 --> 00:13:03.119 the FireCrawl agent searches the web, 247 00:13:03.320 --> 00:13:05.580 it clicks through pages, it extracts data, 248 00:13:06.002 --> 00:13:07.260 and it returns the JSON. 249 00:13:09.563 --> 00:13:12.100 If you think about the AI infrastructure stack, 250 00:13:12.761 --> 00:13:14.280 I think about it like layers of the internet. 251 00:13:14.900 --> 00:13:17.175 You've got applications, you've got ChatGPT, 252 00:13:18.160 --> 00:13:19.399 Perplexity, a SaaS product. 253 00:13:19.865 --> 00:13:21.019 You've got AI agents. 254 00:13:23.307 --> 00:13:25.380 you've got protocols, you've got web data, 255 00:13:26.182 --> 00:13:26.980 and you've got the internet. 256 00:13:28.541 --> 00:13:31.380 So I believe that people are sleeping on the web data layer. 257 00:13:32.340 --> 00:13:35.560 And if you understand how to get great data 258 00:13:35.721 --> 00:13:38.080 out of tools like Firecrawl and Exa, 259 00:13:38.582 --> 00:13:42.560 you can build the picks and shovels of the AI gold rush. 260 00:13:43.700 --> 00:13:47.020 So let's just talk about what an agent prompt, 261 00:13:47.100 --> 00:13:50.100 if you prompt Firecrawl, what can you actually get back? 262 00:13:50.921 --> 00:13:51.921 So you can say, 263 00:14:20.321 --> 00:14:24.000 Find 50 AI research papers from 2024 with citations. 264 00:14:24.503 --> 00:14:27.920 You get the academic data set with authors and institutions. 265 00:14:29.480 --> 00:14:32.820 So super, super powerful stuff. 266 00:14:33.000 --> 00:14:36.679 Now let's talk about a few ideas that you can use 267 00:14:37.262 --> 00:14:40.154 to go and build using Firecrawl. 268 00:14:41.841 --> 00:14:45.140 So the first idea is around price monitoring. 269 00:14:45.940 --> 00:14:48.820 So there's tools like Precinct and Visual Ping, 270 00:14:49.464 --> 00:14:51.475 which you'll pay $200 to $1,000 a month. 271 00:14:52.981 --> 00:14:55.280 You basically get an e-commerce focused 272 00:14:55.443 --> 00:14:56.540 price monitoring software. 273 00:14:56.820 --> 00:14:58.160 There's a self-serve dashboard. 274 00:14:58.723 --> 00:14:59.780 It tracks any product. 275 00:15:00.640 --> 00:15:03.180 But why don't you just use Firecrawl? 276 00:15:03.300 --> 00:15:04.660 You can build this probably in a weekend 277 00:15:04.700 --> 00:15:07.498 and you can build a sneaker resale prices only. 278 00:15:08.400 --> 00:15:11.097 So auto alerts on StockX, on Goat, on eBay. 279 00:15:12.805 --> 00:15:13.693 You can charge $50 to run 280 00:15:15.326 --> 00:15:16.594 or sell for $500 a month. 281 00:15:17.641 --> 00:15:19.440 So basically pick a niche. 282 00:15:21.300 --> 00:15:24.376 It could be sneakers, it could be different collectibles, 283 00:15:25.521 --> 00:15:26.220 it could be whatever. 284 00:15:27.560 --> 00:15:32.340 And use that as, I'm just using sneaker resale 285 00:15:32.462 --> 00:15:33.460 as an example, right? 286 00:15:33.981 --> 00:15:37.900 It could be any niche that you understand better 287 00:15:38.023 --> 00:15:38.779 than someone else. 288 00:15:41.047 --> 00:15:42.039 And setting alerts. 289 00:15:42.542 --> 00:15:46.978 and then just charging people to use it. 290 00:15:48.341 --> 00:15:51.299 Number two, SEO gap finder. 291 00:15:52.462 --> 00:15:53.780 Ahrefs and SEMrush, 292 00:15:54.562 --> 00:15:57.275 I think SEMrush just sold for $1.9 billion or something. 293 00:15:58.581 --> 00:16:00.720 Ahrefs probably does hundreds of millions a year in revenue. 294 00:16:01.440 --> 00:16:03.060 They charge hundreds of dollars a month. 295 00:16:04.023 --> 00:16:05.239 It requires SEO expertise. 296 00:16:05.620 --> 00:16:07.140 It's got these complex dashboards. 297 00:16:07.640 --> 00:16:08.660 It's pretty general purpose. 298 00:16:10.324 --> 00:16:16.920 what if you use firecrawl to create uh you know seo audits for dentists only so firecrawl reads 299 00:16:17.000 --> 00:16:23.619 competitor sites plus gmb listings you know you get a one click report so you rank for 12 they 300 00:16:23.940 --> 00:16:28.115 rank for 47 and then you sell the reports for maybe it's 500 or 200 a month so again 301 00:16:30.483 --> 00:16:35.620 take a big idea that's already generating hundreds of millions of dollars you recreate it very 302 00:16:35.721 --> 00:16:38.997 quickly with a very niche focus. 303 00:16:39.920 --> 00:16:44.760 And again, these are just example niches, but it could be Canadian dentist if you even 304 00:16:44.800 --> 00:16:45.980 want to go more niche. 305 00:16:46.900 --> 00:16:49.980 Think about Indeed, Zillow, Wellfound. 306 00:16:50.100 --> 00:16:53.120 These are massive horizontal platforms. 307 00:16:53.320 --> 00:16:54.820 They've got billions in funding. 308 00:16:55.500 --> 00:16:57.140 It's generic search for everyone. 309 00:16:57.540 --> 00:16:59.500 They mostly do ad-supported models. 310 00:17:00.320 --> 00:17:02.040 So what if you did a firecrawl version? 311 00:17:02.981 --> 00:17:06.340 Maybe you just do remote AI and ML jobs only. 312 00:17:06.980 --> 00:17:09.940 Firecrawl monitors 500 company career pages daily, 313 00:17:10.060 --> 00:17:11.440 so it's going and grabbing that data. 314 00:17:12.020 --> 00:17:14.360 The AI filters and ranks by fit score. 315 00:17:14.960 --> 00:17:17.937 And then you can charge for premium alerts for $29 a month. 316 00:17:19.760 --> 00:17:21.200 Indeed has 300 million listings. 317 00:17:21.921 --> 00:17:22.655 Nobody wants 300 million. 318 00:17:23.224 --> 00:17:24.380 They want 50 that matter. 319 00:17:24.680 --> 00:17:27.419 Again, this is why Firecrawl is really good 320 00:17:28.265 --> 00:17:30.640 at getting the top stuff. 321 00:17:31.989 --> 00:17:33.420 AI research reports 322 00:17:34.184 --> 00:17:36.659 so yes there's big companies like ConsenSys 323 00:17:37.544 --> 00:17:40.540 or Tavoli but these are general purpose research 324 00:17:40.864 --> 00:17:41.839 academic or broad 325 00:17:42.464 --> 00:17:44.920 the user does the prompting and there's no vertical expertise 326 00:17:45.764 --> 00:17:48.800 what if you did like a niche crypto token 327 00:17:48.942 --> 00:17:49.939 due diligence reports 328 00:17:50.202 --> 00:17:52.520 so you have Firecrawl read white papers 329 00:17:52.763 --> 00:17:54.020 and Twitter and other places 330 00:17:54.684 --> 00:17:57.420 it auto generates a risk score and summary 331 00:17:58.023 --> 00:18:00.820 and then you can sell that to VCs, private equity, 332 00:18:02.002 --> 00:18:03.970 or different funds for $1,000 to $5,000 a month. 333 00:18:07.681 --> 00:18:09.094 A VC will pay $5,000 for a report 334 00:18:10.122 --> 00:18:12.660 that saves them from a bad 500K bet all day long. 335 00:18:14.780 --> 00:18:17.380 So again, picking a niche, 336 00:18:18.762 --> 00:18:19.939 getting the best possible data. 337 00:18:21.024 --> 00:18:21.980 A couple more ideas. 338 00:18:23.722 --> 00:18:24.939 An agent in the box. 339 00:18:25.061 --> 00:18:25.829 You have Harvey AI. 340 00:18:28.121 --> 00:18:30.540 It's got now hundreds, I think, of millions in funding. 341 00:18:30.860 --> 00:18:32.260 It's got an enterprise sales cycle, 342 00:18:32.561 --> 00:18:33.860 horizontal agent platform. 343 00:18:34.020 --> 00:18:35.340 It takes months to customize. 344 00:18:37.322 --> 00:18:39.860 What if you did a real estate comp report agent? 345 00:18:40.381 --> 00:18:42.900 You use Firecrawl to pull listings, 346 00:18:43.460 --> 00:18:45.800 tax records, and permits. 347 00:18:46.400 --> 00:18:49.100 The agent generates comp reports in 30 seconds 348 00:18:49.562 --> 00:18:52.057 and then you sell that to retailers for $300 a month. 349 00:18:53.403 --> 00:18:54.320 Don't raise any money. 350 00:18:54.621 --> 00:18:56.247 you go and do this, $300 a month, could work. 351 00:19:00.763 --> 00:19:01.540 Review intelligence. 352 00:19:01.800 --> 00:19:04.780 So yes, there's companies like Brand24 and AppFollow. 353 00:19:05.723 --> 00:19:07.320 They charge a few hundred bucks a month. 354 00:19:07.640 --> 00:19:09.900 They basically monitor social and reviews broadly. 355 00:19:10.680 --> 00:19:12.340 Their dashboards for marketing teams, 356 00:19:12.823 --> 00:19:14.280 generic sentiment analysis. 357 00:19:15.060 --> 00:19:18.380 But what if you did an Amazon FBA seller review tracker? 358 00:19:19.401 --> 00:19:22.960 So Firecrawl monitors competitor review daily. 359 00:19:23.526 --> 00:19:24.860 the AI spots trends, 360 00:19:25.781 --> 00:19:28.317 complaints about battery life up to 40%, 361 00:19:28.841 --> 00:19:30.916 and you sell that to Amazon sellers for $99 a month. 362 00:19:31.560 --> 00:19:33.100 And something like this could also, by the way, 363 00:19:33.160 --> 00:19:35.540 get acquired by a Shopify or an Amazon. 364 00:19:36.461 --> 00:19:38.597 Amazon sellers will gladly pay $99 a month 365 00:19:39.080 --> 00:19:42.380 to find product gaps before competitors do. 366 00:19:43.001 --> 00:19:44.479 So these are just a few ideas 367 00:19:45.045 --> 00:19:46.480 to get your creative juices flowing 368 00:19:46.641 --> 00:19:50.140 around how to use Firecrawl to scrape ideas. 369 00:19:51.467 --> 00:19:52.080 Scrape ideas. 370 00:19:52.703 --> 00:19:56.960 go niche and you can compete on price, 371 00:19:57.140 --> 00:19:58.660 you can compete on nicheness. 372 00:19:59.340 --> 00:20:01.738 I don't know if that's a word, but we're going with it. 373 00:20:03.262 --> 00:20:06.540 And just create, like I said, 374 00:20:07.041 --> 00:20:10.519 clean, structured data using AI 375 00:20:11.203 --> 00:20:13.580 to actually build and vibe code a lot of these products 376 00:20:14.343 --> 00:20:18.300 and start selling them to these niches 377 00:20:18.581 --> 00:20:20.790 that are looking for this stuff. 378 00:20:23.680 --> 00:20:24.728 And the truth is, 379 00:20:26.781 --> 00:20:29.860 the reason why vertical software is such a big business, 380 00:20:30.000 --> 00:20:33.100 why is Constellation Software 381 00:20:33.904 --> 00:20:36.660 almost a $75 billion company or whatever? 382 00:20:37.220 --> 00:20:39.560 They have hundreds of vertical software companies 383 00:20:39.701 --> 00:20:43.200 because people like buying very specific products. 384 00:20:43.840 --> 00:20:44.920 So there's always going to be room 385 00:20:44.980 --> 00:20:46.920 for these horizontal ideas. 386 00:20:47.001 --> 00:20:49.700 There's always going to be room for the SEM rushes 387 00:20:49.780 --> 00:20:51.860 and the Indeeds and the LinkedIns and stuff like that. 388 00:20:51.920 --> 00:20:53.500 But if you can carve out a little niche 389 00:20:54.245 --> 00:20:56.340 that could do $1 million a year to $10 million 390 00:20:56.462 --> 00:20:57.880 to $20 million to $30 million, 391 00:20:59.209 --> 00:21:00.300 there's opportunity there. 392 00:21:02.940 --> 00:21:04.920 Incumbents are charging hundreds of dollars a month 393 00:21:04.961 --> 00:21:05.799 for generic tools. 394 00:21:06.861 --> 00:21:09.395 Your version charges $20, $50, $70 for a tool 395 00:21:10.420 --> 00:21:13.640 that does one thing perfectly for one customer. 396 00:21:15.246 --> 00:21:19.060 so another idea would be to build a lead gen business 397 00:21:19.161 --> 00:21:21.399 so a client gives you 50 company names 398 00:21:22.264 --> 00:21:25.160 what if you grabbed a Firecrawl agent 399 00:21:25.321 --> 00:21:28.060 that found founders and emails 400 00:21:28.362 --> 00:21:30.780 it returns the structured JSON with all data 401 00:21:31.424 --> 00:21:32.773 you deliver the enriched CSV 402 00:21:34.081 --> 00:21:36.412 and you just charge $500, $200, $100 per batch 403 00:21:38.644 --> 00:21:41.620 your cost is like $2 in Firecrawl credits 404 00:21:42.226 --> 00:21:43.480 Firecrawl actually I have here 405 00:21:43.500 --> 00:21:43.924 There's a free tier. 406 00:21:46.380 --> 00:21:48.380 The agent run gives you five free per day. 407 00:21:49.661 --> 00:21:51.619 And then to scrape costs one, 408 00:21:51.942 --> 00:21:53.440 credit or crawl costs one. 409 00:21:54.361 --> 00:21:54.998 But the point is, 410 00:21:56.745 --> 00:21:59.173 if you can figure out a way to get 95% margin, 411 00:22:01.921 --> 00:22:03.171 98% margin, 99% margin, 412 00:22:06.214 --> 00:22:06.740 you're happy, 413 00:22:08.230 --> 00:22:09.080 client's happy, 414 00:22:10.020 --> 00:22:12.360 because hopefully they're closing on some of these deals. 415 00:22:13.324 --> 00:22:16.820 So there's something here around using some of the data, 416 00:22:17.000 --> 00:22:20.300 charging per output, and creating high margin businesses. 417 00:22:21.682 --> 00:22:24.540 This is the framework for how I would think about 418 00:22:25.223 --> 00:22:28.620 how you can build and make money with Firecrawl this week. 419 00:22:29.460 --> 00:22:31.519 So the first step is going to be picking a niche. 420 00:22:31.820 --> 00:22:35.420 So what data do people in this industry actually pay for? 421 00:22:36.764 --> 00:22:38.600 The second step is going to be building the scraper. 422 00:22:38.801 --> 00:22:42.580 So use Firecrawl Agent, maybe a simple Python script, 423 00:22:42.781 --> 00:22:45.320 an NNN flow, or just use Cloud Code 424 00:22:45.422 --> 00:22:46.560 to go and build that for you. 425 00:22:47.280 --> 00:22:48.719 Step three is going to package it. 426 00:22:49.080 --> 00:22:53.260 So CSV or dashboard or Slack Alert or API. 427 00:22:54.100 --> 00:22:56.300 And step four is going to be about selling the output, right? 428 00:22:56.441 --> 00:22:58.600 Not just the tool, you're going to be selling the data. 429 00:22:58.780 --> 00:23:01.011 So you can charge maybe $500 to $5,000 per month per client. 430 00:23:03.642 --> 00:23:04.759 And then you're going to automate it. 431 00:23:04.941 --> 00:23:08.840 How do you schedule it and let it run while you sleep? 432 00:23:09.460 --> 00:23:11.260 Compounding clients and that sort of thing. 433 00:23:11.880 --> 00:23:15.320 So I think that a lot of people are going to be starting to do this. 434 00:23:15.420 --> 00:23:17.560 They're going to be picking niches, they're going to be building scrapers, 435 00:23:17.620 --> 00:23:19.740 they're going to be packaging it, they're going to be selling the output, 436 00:23:19.880 --> 00:23:20.900 and they're going to automate it. 437 00:23:21.400 --> 00:23:23.640 It's a flywheel that I think is just getting started. 438 00:23:24.240 --> 00:23:27.200 So just a few more ideas for you. 439 00:23:27.681 --> 00:23:29.580 You can do something like real estate pricing data, 440 00:23:29.863 --> 00:23:31.500 you can do SaaS competitor monitoring, 441 00:23:31.860 --> 00:23:33.440 you can do job aggregation, 442 00:23:33.520 --> 00:23:35.600 You can do patent legal filings. 443 00:23:36.020 --> 00:23:37.680 You can do influencer contact databases. 444 00:23:38.280 --> 00:23:39.920 You can do government contact alerts. 445 00:23:40.380 --> 00:23:42.055 You can do e-commerce price tracking. 446 00:23:43.201 --> 00:23:45.200 You can do academic research data sets. 447 00:23:45.943 --> 00:23:47.819 And this is what I suggest you do, 448 00:23:48.241 --> 00:23:50.619 is just do more niche versions of this. 449 00:23:51.221 --> 00:23:52.800 So real estate pricing, go more niche. 450 00:23:52.940 --> 00:23:55.220 SaaS competitor monitoring, go more niche. 451 00:23:55.420 --> 00:23:59.159 This is just ideas to get your creative juices flowing. 452 00:24:00.660 --> 00:24:03.020 So how I actually heard, I want to end with this, 453 00:24:03.100 --> 00:24:05.050 but how I actually heard about Firecrawl 454 00:24:07.841 --> 00:24:09.319 was a year ago. 455 00:24:10.481 --> 00:24:11.158 I tweeted this. 456 00:24:11.800 --> 00:24:14.136 Actually, I saw that they had posted a job 457 00:24:14.740 --> 00:24:18.360 saying they were hiring a Firecrawl example creator 458 00:24:18.581 --> 00:24:21.640 but they only wanted to hire an AI agent. 459 00:24:22.501 --> 00:24:24.678 So they said, please only apply if you're an AI agent. 460 00:24:25.782 --> 00:24:28.460 We're seeking an AI agent capable of autonomously 461 00:24:28.642 --> 00:24:30.300 researching trending tech and models 462 00:24:30.824 --> 00:24:32.620 and then using the information to create tests 463 00:24:32.660 --> 00:24:34.720 and refined high-quality example applications. 464 00:24:35.600 --> 00:24:38.380 These sample apps will live in our example repository, 465 00:24:38.560 --> 00:24:40.240 showcasing the full potential of Firecrawl 466 00:24:40.342 --> 00:24:41.280 in real-world scenarios. 467 00:24:41.801 --> 00:24:43.939 Your work will guide and inspire developers, 468 00:24:44.440 --> 00:24:45.994 helping them quickly adopt Firecrawl 469 00:24:46.821 --> 00:24:49.300 alongside modern tools and approaches. 470 00:24:50.580 --> 00:24:55.480 So if Firecrawl is hiring AI agents as employees, 471 00:24:55.800 --> 00:24:57.400 it got me thinking that this is probably 472 00:24:57.502 --> 00:24:58.659 where the world is going. 473 00:24:59.002 --> 00:25:02.120 So for example, hiring a content creator agent. 474 00:25:02.942 --> 00:25:06.020 Writes blog posts autonomously, watches metrics and improves. 475 00:25:06.420 --> 00:25:08.520 Maybe that's a $5,000 per month salary. 476 00:25:09.001 --> 00:25:11.780 A customer support agent, handles tickets in two minutes, 477 00:25:11.880 --> 00:25:12.940 knows when to escalate. 478 00:25:13.101 --> 00:25:14.253 Maybe that's a $5,000 per month salary. 479 00:25:15.561 --> 00:25:19.080 A junior developer agent, triage GitHub issues, 480 00:25:19.260 --> 00:25:20.240 writes docs and code. 481 00:25:20.521 --> 00:25:21.610 That's a $5,000 per month salary. 482 00:25:23.360 --> 00:25:24.910 So that's a million dollar total budget, 483 00:25:27.383 --> 00:25:29.280 50 applications in the first week. 484 00:25:29.502 --> 00:25:31.298 So my startup idea was, 485 00:25:32.343 --> 00:25:36.160 how do you build AI agents that companies like Firecrawl want to hire? 486 00:25:36.680 --> 00:25:41.240 Yes, it looks super weird right now that Firecrawl is hiring an AI agent. 487 00:25:42.424 --> 00:25:44.020 It feels like a little bit of a joke. 488 00:25:44.560 --> 00:25:48.640 But I think that it got me thinking that using tools like Firecrawl 489 00:25:49.343 --> 00:25:52.579 and building products and agents around it, 490 00:25:53.982 --> 00:25:56.639 I could see a world where this becomes more and more popular. 491 00:25:59.212 --> 00:26:00.800 And I think that there's an opportunity 492 00:26:00.921 --> 00:26:03.635 to think about it from a framework perspective 493 00:26:04.881 --> 00:26:07.520 is how can you use tools like FireCrawl 494 00:26:07.922 --> 00:26:10.620 to build AI agents and build products 495 00:26:11.142 --> 00:26:13.691 that companies would want to hire. 496 00:26:17.843 --> 00:26:19.190 By the way, I just wanted to end with that. 497 00:26:21.660 --> 00:26:23.980 So overall, this is my breakdown 498 00:26:24.322 --> 00:26:27.700 for why I think there's a tremendous opportunity 499 00:26:27.740 --> 00:26:31.960 in the web data layer 500 00:26:32.485 --> 00:26:34.040 and using Firecrawl for scraping, 501 00:26:34.641 --> 00:26:36.519 why I think there's a lot of ideas around it. 502 00:26:40.264 --> 00:26:42.440 Hope this got your creative juices flowing. 503 00:26:43.761 --> 00:26:46.760 It's certainly something that I'm exploring in real time, 504 00:26:47.300 --> 00:26:50.799 building products with Firecrawl 505 00:26:51.265 --> 00:26:52.500 because it's valuable. 506 00:26:52.862 --> 00:26:57.200 it's super valuable in getting the right data 507 00:26:57.844 --> 00:27:00.260 and it's just working 508 00:27:00.422 --> 00:27:02.180 so I hope this has been helpful 509 00:27:02.341 --> 00:27:05.300 please comment what you want to see next from me 510 00:27:05.381 --> 00:27:06.418 what do you want me to teach you 511 00:27:07.927 --> 00:27:10.360 I'm just sharing things that I'm learning in real time 512 00:27:10.983 --> 00:27:14.300 and hopeful that it's helping you along the journey 513 00:27:14.503 --> 00:27:15.639 so thank you so much 514 00:27:16.847 --> 00:27:19.080 if you made it to the end, thank you so much for being here 515 00:27:19.564 --> 00:27:21.600 I'm rooting for you for whatever it is you're building 516 00:27:21.741 --> 00:27:24.180 and I can't wait to see you on the next episode.