Build Better Apps On Easy Mode with Doc Willams

🎁  Download Part 1 Here: https://go.n8n.io/app-creation-workflow
🎁  Download Part 2 Here: https://go.n8n.io/app-creation-workflow-prompts

Doc Williams shows you how to build a complete SaaS application that scrapes app store reviews, analyzes them with AI, and helps you identify million-dollar opportunities in underserved markets.

The System:
This workflow automatically scrapes app reviews, lets users chat with the data using Pine Cone's new assistant node, and even generates Product Requirement Documents (PRDs) to help you build better apps. The entire system uses Fire Crawl for scraping, n8n for automation, Pine Cone for AI chat, Lovable for the front-end, and Stripe for payments.

Tech Stack:

n8n (automation & workflows)
Pine Cone Assistant (AI chat with your data)
Lovable (front-end development)
Supabase (database)
Fire Crawl (review scraping)
Stripe (payments)
OpenAI (language model)

What You'll Learn:
• How to scrape and analyze app store reviews to find profitable gaps in the market
• Setting up Pine Cone's Assistant Node with n8n (no complicated vectorization needed!)
• Embedding AI chat directly into your SaaS application
• Saving chat history to a database with Supabase
• Building the entire front-end with Lovable AI
• Adding Stripe payments to monetize your app
• Real example: Oil Field Calendar app making $50k+/month

0:00 - Introduction & Overview
0:22 - What You'll Learn Today
1:00 - The App Review System Demo
2:16 - How to Find Profitable App Opportunities
3:52 - Live Demo: Oil Field Calendar Example
5:30 - Using AI Chat with Pine Cone
9:30 - Embedding the Chat Feature in Your SaaS
10:17 - Tech Stack Overview
11:57 - Inside the n8n Workflow
15:00 - Understanding Pine Cone Assistant Node
17:30 - Top K and Snippet Size Explained
19:47 - Data Shape and Configuration Tips
22:09 - Metadata Filters Setup
22:56 - Saving Chat History to Database
24:26 - Generating Product Requirements (PRDs)
26:40 - Full Tech Stack Breakdown
29:52 - Setting Up with Lovable & Stripe
30:30 - Template & Prompt Availability
31:14 - Future Opportunities & Market Ideas
33:26 - Closing Thoughts & Where to Connect

Get 30% Off n8n Cloud Starter or Pro Plans!
Want to get started with n8n? Visit n8n.io/pricing and use code 2025-N8N-PODCAST-729C416E at checkout for 30% off your first month or year.

Transcript

SPEAKER_00 0:00

Basically, they just put it on easy mode right out of the box. It works. Boom. I've got based on your reviews, here are the top key areas to focus on with making your own application better. No must, no fuss. All right here. We're going to be talking about the small app that I've built, a micro app that I use for myself and for my clients, and then how I'm using N8N in the new Pinecone assistant node to do that.

SPEAKER_01 0:22

So this is all the heavy lifting that you don't have to do because it's automatically processing it. Doc Williams built a system that helps you build something better. Today, each daring is the exact system he built to do this. It scrapes the reviews of apps in the store, it chats with the data using Pinecone, and then it spins up a full stats app with lovable, in it in and stripe. The full workflow and the lovable prompt to build your own system is inside the description down below. Let's get into it. Doc Williams, thanks for coming on the podcast, brother. Excited to have you, man. Appreciate it. And so my main question for you is this What are we going to be learning today? Hey, man.

SPEAKER_00 1:00

Well, first of all, thanks so much for having me here. What we're going to be talking about is how to use basically the chat features and how to save things to a database using N8N and Pine Cone and lovable. So we're going to be talking about the small app that I've built, a micro app that I use for myself and for my clients, and then how I'm using N8N in the new Pine Cone Assistant node to do that.

SPEAKER_01 1:24

Love it. Let's dive in it, man. So if you feel ready, share your screen and let's look at it.

SPEAKER_00 1:29

Okay, good times. So, and what I'll do is also when I'm presenting, there is going to be somewhere, somewhere with when this video comes out, you'll have the workflow and be able to see all these things. So we'll go uh step by step with that.

SPEAKER_01 1:43

Yeah, if you look down the link below, there'll probably be in the description of the YouTube video or LinkedIn. There'll be a place to click it, download it, and get it for yourself.

SPEAKER_00 1:49

So good times, good times. So let's go through an overview of like what it is, what we're doing, and all those things. And then we're gonna go step by step. And if I miss anything or if you have questions, please let me know. So this is basically the workflow for the AI chat agent with a database history and logging all those things. So bottom line, you're gonna be able to have this on your site, on your application for people to start chatting with the chatbot. And then they're gonna be you're gonna be able to save all of that information in a database. And we have this using that in uh lovable right here and using Superbase. So, bottom line, what we're gonna do is let's zoom in for a moment. We're gonna have a chat input. What we're gonna do here is we're having it embedded on the site. From there, we're having the AI agent that's going to be with OpenAI and the chat uh model, and then also the Pine Cone assistant node here. And then what we're doing is posting the response so we're having all of the save messages go right back into the database so we can save those at that information. And then we're going to also have the formatting answer so that way we can have it in the chat and we can have it in our database. So when the person responds, there won't be like, hey, where's the response? And like, what happened? Goes into the database, you'll be able to chat with it and all of those different things. Um, we can go into a little bit more, but basically, this is an idea from uh Greg Eisenberg came out with a post on X not too long ago talking about, hey, there's a way to make about$10 million in the next year. And one way would be looking at all of the different applications that are in the App Store, look at which ones are going well, which ones aren't doing well, and then being able to learn from their mistakes and then launch your own. So what we can do is kind of use this as an example, talk about, you know, see it in live practice and go from there if you want, man.

SPEAKER_01 3:50

Great. Yeah, let's dive into it. So it sounds like what we're doing is we're looking to arbitrage and see what apps are valuable inside the app store. And can we build something better uh using these no code solutions so that we can say, okay, there's a gap in the market, here's an opportunity for me to build something better. And so this the system, it sounds like with Pine Cone, you have a because before it's a little complicated to be able to vectorize your memory, right? There's some, you know, you need to know how to chunk it, there's embedding issues and things like that. So small versus large. So it sounds like this handles a lot of that heavy lifting for us.

SPEAKER_00 4:21

Absolutely, absolutely. Sometimes it was just stressful. It was like Pinecone had so many great things, but the testing process, understanding all of those things, like basically they just put it on easy mode, right out of the box. It works in all of those things. So we'll we'll have this right here. And basically, what I did was I I had the application. This was great. Let's uh let's sign out for a moment so you can see what's going on here and kind of go right into it. So let's refresh right here. And so right now we can have the app review exporter. If we look right here where it says like pro, it's like, oh, before you can use AI chat, you've got to pay for it. So right here we we have it right behind that paywall. So they're like, okay, we can test it out, see if we like the app, and uh, if we want more support, we can get that chat feature that we're we're charging for.

SPEAKER_01 5:16

So And I just want to clarify. So this is essentially a mini SaaS app that you built to service the need of the difficulty of finding reviews online. So then you've hooked this up to be able to make it really easy to find those reviews, and then you just put a paywall in front of that so that people that want to use the service that won't have to build it out can just sign up easily. And that's kind of the Greg Eisenberg philosophy of making something that is difficult, easy, and then turning it into your own SaaS app.

SPEAKER_00 5:40

Exactly. Yeah, you said better than me. That's exactly what's happening right here. That's right. And you see right here, it's like, okay, that's great. Uh, 90 days, that's good. But hey, I want all time. And you notice right here, there's a couple different lock buttons. It's like, oh man, I want that. Um, oh man, I've got to pay, right? So it's still providing value. You're able to see it. And depending on the type of app, um, how how big it is, you're gonna be getting tons of, you know, it can vary how many reviews you're gonna be getting. Right here, we're just actually looking at something that is very profitable, which is just wild to me. It's called oil field calendar, and it's for technicians that are needing to track their schedule, working on a rig and everything. People, uh, this app is making more than 50,000 a month, just about. It's just like, I cannot believe these things are out there. Um, so this was an example of like, okay, this is out there, it's already profitable. Can we make it a little bit better? And so, right here, okay, that's great. I see that I I have some reviews, but I want more. I'm gonna sign in with an account that already has uh the pro features. I'm gonna search for it again, and then we're gonna see how many more reviews that we could possibly get. Okay, so we have 90 days. Let's make it to all time. Okay, so we have 147 reviews. Let's just look at everything for four stars and less. So we have 53 right here, and now we're getting some data of like what's not working well with the application. And so a couple of things. We can analyze it and we can have a couple, you know, a little bit of a details that we have uh with with everything that we have hooked up. But also, too, as it's working on analyzing, I can export it so I can have the CSV, and I'm gonna show you in a moment that I can chat with it in a second. And so, okay, I can analyze that's quick for a second. You'll see that it's thinking about its life. And what we're doing here is we're just using, you know, lovable and using lovable AI, and it's connected. We can see some things like okay, top complaints, but I want to go a little bit deeper. So then I can pop open the app review AI. And so this is where we're using the chat input from N8N. We're embedding it right here on the page, and we're now connecting it with Pinecone. And so I'm gonna attach the CSV that I just downloaded. Okay. And then what I'm gonna do is just say, like review the attach CSV file and tell me the top three different things that I should be looking for to make this application better because I want to create my own. And so I'm giving it a little bit more, but again, it's going to be allowing me to chat directly with what I'm uploading right there. So it's gonna be thinking about its life for a moment, all of those things. We'll see.

SPEAKER_01 8:35

What I like about this too is the fact that you've embedded the N8N chat feature into this actual SaaS application. So you have the advantages of N8N going into the system, but it's on a page. So you don't need to bring someone to N8N or a different website or anything else. It's all integrated in there.

SPEAKER_00 8:50

Mm-hmm. Exactly. And then right here, boom. I've got based on your reviews, here are the top key areas to focus on with making your own application better. I can use that now. I'm like, okay, great. Now I can go about what do I want to do? I want to take it somewhere else, I can do those things, but now I've got it curated directly with what I have right here. It's hitting using Pine Cone, no muss, no fuss. All right here. Um, and then we can kind of talk about the what it's saving in the back end too, and how I could be using that too, either as you know, an admin seeing all the history or going from there. I'll stop there for a moment though.

SPEAKER_01 9:26

Yeah, let's take a second, let's look at the workflow. There's some features on here I do want to talk about, but before we get into that, let's dive into the in it inside, see what's going on in the back end, and then we can have some additional conversations.

SPEAKER_00 9:38

Okay, cool. So what we're seeing right here, um, let's see, let's do this. So when we're looking at it, you tell me where you want to go. I'll stop. Let's stop.

SPEAKER_01 9:50

Let's look at the let's look at the execution history. Let's see what came through the chat side, and we can take it like one step at a time from the execution all the way to the the output. I love it. And this, oh, go ahead, go ahead.

SPEAKER_00 10:02

I don't want to.

SPEAKER_01 10:03

And in terms of setting this, this, this up, the the downloadable template. So this is gonna have all those elements in there. Is there any what additional credentials will people need to set this up? What what API? I'm assuming Pinecone for sure, but is there anything else that they need to set up to make this function?

SPEAKER_00 10:17

Yeah, great question. So with this one, you're gonna have to use OpenAI. So if you have that one, you're gonna be able to have that all set up. And then you're going to be using Pinecone. So you can create an account. You can start for free. And then once you create an account, it'll ask you to create an API key. You have that, you copy it over, make sure you save it in a safe, safe, secure place, and then you'll bring it over and you can have that all together. Um, I I just personally use OpenAI, but um, you know, if you want to use a different uh chat model or anything like that, you're good to go. You can just change it out right here. Um, the that's pretty much it. The other thing is we're just going to be saving it to the database. I'm asking it. I'm using lovable and superbase. So I'm going through and chatting, hey, what's the webhook I should use? So then when it's done, it's doing a post request, it's sending the chat over to that database and everything like that. So whatever database you plan to use, you could use Superbase or other things, just a place where it's going to be saving all that information for you to, you know, review that that information later on.

SPEAKER_01 11:21

Yeah. And two notes on that. I know we're using the intelligence model of OpenAI, but if you wanted flexibility, we could always switch that out to open router. Open router would allow us to then pick and choose any other models that we want. So that's a great node. And then also I'm seeing here that there's uh looks like there's an error or some sort of uh warning we have sign on there. Yeah.

SPEAKER_00 11:39

Yeah, and this is because we're in the template. I can switch over to what we just had. Let me switch over so we can kind of look at the execution and everything. So let's look at the this one right here. So I was over here and basically I built on to the different pine cone workflow. This is a little bit different, but basically, yeah. So now we can see the execution and everything like that if we want to go. But that's why that we had that error right there. Yeah.

SPEAKER_01 12:04

Got it. It's just it's the template error. You just need to put in the credentials for error for pine cone.

SPEAKER_00 12:09

Yeah, yeah, yeah. All right. Let's see. So we have this section here. Um, okay, I'll stop here. Should we go over to execution and kind of? Okay, let's see. So let me actually, I'm not usually over here with execution. So please, Dylan, you might have to walk me through if I'm doing something wrong or if I'm not um doing the please.

SPEAKER_01 12:35

No, this is great. I mean, what you can do is you can open this, you can uh what the section number two says chat with your docs. You can open that up and we can see the data moving through there. If we want to make any adjustments, you can always hit copy to editor. It's a great function. But we can see here um we're we're getting in. It looks like this is the CSV that you've extracted and sent through the chat uh widget that we have to then be able to then pass that that file over. All right, okay, cool.

SPEAKER_00 13:01

And then here. And then let me zoom in really quick. Sorry.

SPEAKER_01 13:07

I'm sure people appreciate being able to view it. This is great.

unknown 13:11

Yeah.

SPEAKER_00 13:14

And then we have it right here from the input, right here. Oh, I'll I'm just here. I'm just here. You tell me where talking too, talking too much.

SPEAKER_01 13:21

No, and so we see it's being processed through here. The you see the uh the we're using the the JSON chat input, which is kind of a standard out the gate. It knows to use that. Uh when you ever have it hook it up an AI agent to the uh the chat input. So that's kind of a bit of the template. Then reviews customer. Okay, so we're having those inputs going into there, and then let's see what's happening inside of the uh the pine cone uh node itself. Let's take a look. Okay. Okay, so updating, it's getting the score, you know, the markdown status available. If we scroll all the way down to that, if we scroll down on the responsive side on the right hand side, uh it's already see, it's already got the top case 16, so it's pulling the 16 most relevant results. We see that what's being stored. Okay. Interesting. All right, so this is all the heavy lifting that you don't have to do because it's automatically processing it. Great. Source tag, it's tagging it for you. Beautiful. Alright, look at that.

SPEAKER_00 14:26

And talking to the pine cone team, I was wondering if I should add into the other video that I just did, but basically, a lot of these things I didn't understand um the different snippets, what that meant. And so they were breaking down of all the complications of what it does, which is just fascinating. And sure, just is there any right there?

SPEAKER_01 14:43

Is there any lessons that you got from the pine cone key team that you want to share?

SPEAKER_00 14:47

Yeah, so one of the things was number one, just realize that you have to think about um your data. Every data, like that is different every single time of what you're trying to search for and what the output is. So it's not a one size fits all. You're gonna have to think about as you're going through the different prospects of your data, you're gonna have to keep iterating on it. It's not just going to be like, oh, I I connected the node. This should work right off. Uh, like there's no uh exceptions. Depending on what you're asking for and your use case, you're gonna have to consider that. Um, and that was really insightful when I when I was looking at that. Yeah.

SPEAKER_01 15:31

What adjustments did you have to make to it to have it work for this use case?

SPEAKER_00 15:36

So this is the really funny part. I didn't need to change it at all. I didn't need to change it all. What trip me up actually? I'm gonna I'm gonna be honest with you. Let me show you what I mean by this. And we actually, the the pine cone cone team we're talking about this. Let me do this. Um, so let me come in a little bit more. So the ideas of customizing, there's a section where, like, okay, you need to think about the top K, the snippet size, all these things. So I started going in and really messing around with it. And the more that I because I didn't understand top K, I didn't understand snippet size, out the box, it worked fine until I started going here because I thought, okay, I need to tweak. So that's actually a lot of good insights from the team. They're like, we've already done the heavy lifting to structure it the way it needs to be for a general case, depending on if you needed uh say, for instance, more context or the type of files that it was returning, then you would need to look at changing top K, changing the snippet size and all of those things. So um, yeah, that was that was interesting. Because right here, I was like, oh, I've got to think about all those things for my use case. I thought it was, you know, I was showing me where you would change these things.

SPEAKER_01 16:53

And and also specifically, I'd be curious about the metadata filters because a lot of times uh when you talk about vectorizing data, we want to be able to add some sort of meta tag filters, like whether this is a client or this is a certain project, or you depending on how you're the the data that you're looking to search for, what do you what do you need to adjust to be able to uh this is let's say with the filters?

SPEAKER_00 17:15

Yeah, yeah. So let's do this. I'm gonna do two things right here. So the first thing is uh hello. I'm just gonna say real. So the first thing is they mentioned like just go into the chat and see what's going on when you're bringing everything in, right? So the first one is just like, I'm just testing it out. I'm just gonna say like, hello, testing this out. I want to know more about the pine cone assistant node, right? So it's first just going in, looking at uh the the data right here, and then starting to see when it's saying getting uh context right here, then looking at the content right here and going through this. This part I was like, oh, I I didn't even know this. So this was actually looking at where we're seeing the size, the metadata, looking at what's happening here, and then when you're doing it, it's looking at the return, the return information to see is this matching with what I would be expecting for the output. And so this was something that was super helpful and practical to for me because when uh they were first talking about like Top K in all of these terms, I was like, bro, I don't understand that. I don't know what that means. So this was a really good reflection of like, let's take a step back, let's first just test, and then you start looking at the responses and seeing what you need to tweak. And that's what for me, that was the unlock when they said that.

SPEAKER_01 18:42

Sure. So it sounds like that there's a couple of levels in which you can make some adjustments. You can one make some adjustments for the AI agent. So when the data's coming in, it knows how to filter process. And then the other levels are you know, adjusting the snippets and also the filters and also or the meta tagging uh of those, and then uh the top K results, which is like how many results do you pull back um from the snippets to be able or from the information that vector data store to be able to say, okay, cool. I want five top results, or I want 15 top results, top K top results uh for that. So, okay, that makes sense. Yeah.

SPEAKER_00 19:17

So then could I say one more thing? I'm sorry. Yeah. And one of the things I would like to point out, like, so for me, I wrote down in my notes, there was a couple things like top K. I was like, what in the world? How can I like know what that is? Just so top K, just the only thing we have to think about is it controls the number of search results returned. Like, so I'm like, okay, top K equals that. Okay. Then it was okay, what's the default? 16 snippets. Okay, now I know when I'm going through this, because when I was looking at this, I was like, I don't like how many snippets like this seems like forever. Okay, 16 is the standard. Do I need it less or more? So the more results, it equals the higher token costs, more context.

SPEAKER_01 19:54

So that's good knowing the token cost. And then, yeah, in terms of getting like accurate information, that's where we can get dive deeper into the meta tag filters that would allow us to filter for certain types of informations, whether it's um, you know, uh, if you're gonna have the AI label something as an angry response or something else, you could add that as a filter that then gets uploaded for these are these are negative responses. Or if there's maybe this AI could analyze it, you could have a meta tag filter that would allow you to say this would be uh, for example, feature ideas that could be uh meta tag filter that could get added to this so that you can then say, Hey, I want you to search all the feature ideas of what's possible, and then they'd be easily sorted, and then it would be pulling up the top five, you know, top K or top 16 of the features that it's pulling out of all the reviews. Yeah.

SPEAKER_00 20:41

Yeah. Cool. Um, oh, and then I just have one last uh thing that they mentioned. Uh, so they were like, customization depends, and I I couldn't remember before, it depends on the data shape. And I was like, that's really interesting. In my head, I was like, oh, data shape. And they're like, so it really depends on the structure of what you're putting in there, and the query patterns is gonna depend on what's going on. So there's no universal best practices, but you need to think about the testing and then real data. From there, you can kind of change it. So when I was thinking, oh, the shape of data, that makes sense. When the they're giving us the best out of the box, but depending on what we're trying to input and then output, that's gonna depend on how we're gonna be structuring this, what kind of changes. But the beauty of Pine Cone Assistant app, they've done all the heavy lifting of that configuration. So it's like, I don't have to think about that right to test it right out the box. Now I can go back and you know get that feedback loop and go from.

SPEAKER_01 21:36

And do you know how to do if we're gonna add like meta tag filters into this? Do you know how to do that inside of here?

SPEAKER_00 21:42

Let me check. Let me check. I think it's in my notes. Let's see, that might be. I don't know. I'm seeing in my notes, I'm seeing, let's see, both parameters are helpful. I don't know. I can check though. I can check though, but I don't, I don't think I do. I don't think I I do. But we can check it out. I I might have to do some searching on this.

SPEAKER_01 22:05

Scroll down and see add add custom or add field to see if it's popped up. Yeah, metal filter. There it is.

SPEAKER_00 22:09

Yep, yep, yep, yep.

SPEAKER_01 22:11

Okay.

unknown 22:12

Cool.

SPEAKER_00 22:13

Yep, yep, yep, yep.

SPEAKER_01 22:14

And then you can add that. Then you have the key pair value. So, for example, the key could be feature ideas, or the keys could be angry complaints or you know, whatever it might be. And then that's now being populated inside of that value. So the key pair values, just like any other JSON, is being populated across. That's that's how you're doing it. You can even have the with the magic of those little star icons next to it, um, be able to have the maybe keep the key the same so you don't want it guessing, but then you can have the the values be inputted and have AI process it. Okay, cool. So that's how you send in the the the values.

SPEAKER_00 22:46

Love it. Oh love it, love it. And thank you. Thank you. Cool. Um, yeah, and then we have that, and then if we do you want me to keep going and kind of talk about this one? Yeah, yeah. Let's move on to the the AI agent two. Yeah. So this one is basically it's posting to Superbase and it's going to be adding everything into the database. So for this use case, I want to return in the chat to know where you're going to be going next, but I also want to have uh an admin section where I would have all of the different history. And if I wanted to, I could generate the PRD based off of that. And so this is saving all of the AI history. And again, this was the first iteration. We could also have it like, say, for instance, if you're signed in with the ID that matches, you know, what you just did in the chat, da-da-da-da-da, it would save that history and you could do a PRD. But I wanted to almost be like, could I have a feature that when I'm looking at everyone's like patterns of what they're looking for, could I create like standard PRDs to um say this is what's happening, this is what people are interested in, and kind of going for that idea. So right here it's have oh, yeah.

SPEAKER_01 24:01

And just a note on that, the the PRD, uh, for anybody that's that's listening, it's it's really about what are the requirements for building an application? And so that's a short term that uh that essentially says, okay, I want to I want to build this thing out, which would be great if you're gonna build this out manually, or if you're gonna feed it into AI and say, here's the PRD, use this to design out the application.

SPEAKER_00 24:21

Exactly. And so what we have it right here, it's posting, we have all of this stuff. Um, for for this one, we've done the criteria, we have the URL that it's gonna be hitting and posting all of that information, and the output would be there. And so um, yeah, we have all of this saved data, we can decide to do what do we want from this, make it a PRD and go from here. Um, yeah, that's what we have.

SPEAKER_01 24:46

When you're generating PRD, is it is it based off of like so, for example, let's say you this oil calendar system, if you upload the data into the chat, go chat back and forth with it. When you when you say generate the PRD, is it based off of just one of the reviews, or is it based off of uh holistic reviews or great question?

SPEAKER_00 25:08

So that's where you're having the export of the CSV depending on what you chose. So for mine, I'm always looking for the results, four stars and less, but you could always just do export it on only one star reviews or two stars, and then that will allow you to kind of get a better approach of where you want to go. For me, I find that this is an old trick for like copywriters used to use this for like Amazon reviews back in the day. You would basically use four star and below, usually four stars, four and three star reviews, you get the most passionate fans and the most articulate fans. If you go with a one-star review, a lot of people are like, This sucks, it didn't work for me. But the three and four, they'll write small novellas, they'll talk about what they did, the day it failed, how you ruined their life. Like they go really deep. And so I'm looking for those four and three stars when I'm when I'm exporting. Yeah.

SPEAKER_01 26:04

Makes sense. Yeah, usually with four stars, I would have given it a five, but here's why I didn't it, right? Or yeah, so it makes sense because it's it's the most realistic feedback that you're gonna get. I could also see, too, you know, it's generating PRD, but I I could also see it saying, you know, if you had another button inside of there, say, uh, you know, in it in workflow template. Like what would be if I was gonna feed this into a innate in, what would this look like? And so you could say, generate me uh, you know, a workflow template based on this, and you could feed that back into the system with a one button click that would then give you the back end of it.

SPEAKER_00 26:36

So yeah, yeah, yeah, yeah, yeah. That's great. That's great.

SPEAKER_01 26:39

I I like that idea too. I take it. So then, so then what are you doing with uh with you're you're building out this uh the initial flow on the front end? Uh what did you do to create the lovable app? And then how did you connect? I guess yeah, I guess lovable is the whole system. And did you just embed the chat window and that's all you needed to do inside of lovable, or how did you connect the two systems?

SPEAKER_00 27:00

So a couple different things. Basically, so let's talk about the full tech stack maybe for it with that. Um okay, so with this one, I'm doing lovable on the front end to first scraping the data. I'm using just fire crawl and using that API, bringing that in. And after we have that, the results uh again were just exporting the CSV. The first time I tried it a couple different ways, you can do it a bunch of ways, but basically I fed it. I was like, look at the N8N documentation, look at how to embed this, make this look good. Then basically, it had that. And the reason I wanted it a little bit more custom, I was like, look at how I want it. I also want it to only be shown when they're signed in and it's a pro user. So the first time I was like, hey, show it to everyone. Then I was like, okay, let's do it. And so just a little bit of like back and forth, just talking to Lovable, how I wanted it. And then the other thing is I would screenshot it and I'd be like, this looks good, this doesn't look good. And I would just be like, okay, change these little things. And like the notification when you're not signed in, it will like ping or make different colors to like indicate that hey, there's something that you could be using. So yeah, I just I just gave it the in embed information in the documentation from N8N. Yeah.

SPEAKER_01 28:18

Nice. And I mean, it is the the the modern practice of of AI engineering here is feed the documentation, say sort it out yourself. Wonderful. Now, when you're when you're doing this, you have a payment as well. Is that is that payment via Stripe or how do you hook that up?

SPEAKER_00 28:34

Yep, yep. So I just did Stripe and um I I set it up in you know in Lovable and just said, hey, created a product, created everything right there, and then you know, added a webhook, all those kind of things to for custom. And then just, hey, when they're you know in these conditions, make sure that you tell them to sign in and pay. It does it, and then you come right back onto you know the app and you're good to go.

SPEAKER_01 28:59

Great. Then so you're using Firecrawl, using Stripe, using Pinecone, using Lovable, and you're using N8N. Is there any other pieces of the tech stack that you're using?

SPEAKER_00 29:08

Uh no, no. That's uh that's about it. And you know, and I was also saying, like, I iterated on this a little bit. First, I was like, hey, let's just see how we can start uh, you know, exporting. And then we started talking about like the new chat features and with Pinecone and what's going on with N8N. And, you know, we might add on more things in the future. I think I'm basically just doing a lot of feedback and seeing what people wanted. Um, it it started because on on X, like I said with Greg, with Greg's tweet, and then other developers were like, hey, you don't have like uh Mac reviews. And I was like, Oh, I need to add that. And like, and so I'm just going with the feedback of either different things that we should be adding, or yeah, what what do they want too? And that's that's how we're rolling this thing.

SPEAKER_01 29:54

Great. So when this template becomes available, I mean would be really helpful is maybe inside of the workflow, embedding in the prompt or maybe asking Lovable to create a prompt that would allow them to be able to spin up something similar based on the output that you've created so far so that people could copy and paste that in. I don't know if you put that in there or not.

SPEAKER_00 30:11

Yes, I did. Oh, look at you. Look at head of the game, my friend. But you said it last you're like, get yourself ready. I was like, all right, all right, let me let me get it together. Uh and so yeah, you can uh copy the prompt of the same exact kind of thing, have it good to go, and you can have something similar. You can do it for reviews. I would say also tons of other things. You look at what you can have in the market and you're good to go to get started.

SPEAKER_01 30:39

Yeah, what I love about this is the fact that I mean, the oil rig calendar, I mean, how many apps are in the app stores, right? There's, I mean, a gargantuan amount of opportunities. And the one thing that I I learned from the other podcast guest is the fact that he said the the future is here, it's just unevenly distributed. And so right now there's an opportunity in this space to see, okay, in what markets and what fields like oil rigs or other ones, are there uh some sort of old giants that have had had the market edge just because they've been around and no one's built a better mousetrap. So if you can find these opportunities, you can spin up a system like this and start immediately generating value for those customers and serve them something that they're gonna want and get out to the market. So I think it's great.

SPEAKER_00 31:21

Thanks, man. Yeah, it's it's a it's a great time. Like I'm excited. There's just so many opportunities out there. So yeah, it's great to see.

SPEAKER_01 31:29

What I'd love to see, just playing with this idea a little bit, would be a fun little add-on later on is if you had the generate my market uh marketing content at the back end. So I make the application, great. All right, cool. What would this look like in terms of posts? What would this look like in terms of a website? Can we can we have a can we have a whole business uh spun up off of this based on an idea and say, okay, now that I have my application, what how do I get this out to the masses? And I could I could see building systems for that as well. Love it.

SPEAKER_00 31:55

No, I love that. And um, no, I'm to be continued on that. And then also too, we'll have the video. I just did a video about uh the Pinecone Assistant, like from scratch. So I have the YouTube video attached over there to watch. Uh, you'll have this workflow and you'll have the lovable prompt to get started.

SPEAKER_01 32:12

Fantastic. Is there is there any other feature sets that you'd like to add to this? A wish list and uh concepts. I know anybody that sees this, feel free to comment down below on things that you'd love to see this thing happen with. I'd be able to uh build a uh taller skyscraper with this. Is there any other features that come to mind?

SPEAKER_00 32:28

For this one, not really. I I think you know what? I was looking at basically, I was really inspired. Um sometimes less is best. Like seeing all of these different things, I'm seeing all of these one feature, really simple apps, the simpler, and they're making so much money. So, like to me, I and I even changed the format of like this one because I was looking at other profitable apps and I was like, build me this and have it exactly like this. So we'll see in the future. Um, but I I see different verticals or something like a rinse and repeat I could see for sure. But um, yeah, I got to think more about the futures in the future.

SPEAKER_01 33:06

That'd be a cool use case too, if you found a unique applas, uh, for example, the oil rig calendar, but you knew there were some other similar spaces that could use it that's very successful. Say, great, could you make this, but not for oil rigs, but for real estate or whatever. Name the other industry a vertical that you could you could stamp and repeat for another another season. I know um someone I was working with a while back, they were making like a calculator for macros and that stuff. They're like, oh, but I'd like to do it for vegans. Vegans would be great because they don't have a lot of ability to get protein intake. And so they want to kind of stamp and repeat this in a in a niche specific market. So I could see that as well. What are some underserved markets that this would also add value in? Yeah, no, it's endless, man. It's endless. I love it. Clean and fantastic. So yeah, so we'll we'll make this workflow available uh to the to the masses so you can download it, get access to it. Um, and then again, any comments that you have, uh, feel free to drop it down below. That's how we we learn and grow in this ecosystem together. So uh Doc Williams, it's been awesome, brother. I appreciate you taking the time being on here and uh showing us what you got. Is there anything else you'd like to let people know about? Yeah.

SPEAKER_00 34:11

Yeah, no, I just appreciate being here. It's been awesome. Uh, just the community with N8N and just learning and growing. And the the main thing is what I love about this community, I don't have all the answers, but I can ask someone else and they have an answer. So this has been great. And you know, I appreciate it. If anyone wants to know more or follow the journey of what I'm creating, feel free to uh check out the YouTube channel or drop me a line of what what you're creating. I would love to talk with you and thanks so much for having me, man.

SPEAKER_01 34:40

Awesome, brother. And then what is your YouTube channel so people know?

SPEAKER_00 34:43

Yeah, so it's uh just youtube.com slash at docwilliams D O C Williams. You can find me and I'll be there. Awesome. Much love, my friend.

SPEAKER_01 34:52

Many blessings, and I'll see you on the other side, brother.