Build an AI App with Claude Code in 45 Days (No Coding Experience)
Prefer to watch? Watch on YouTube · Subscribe for new builds · Read the transcript
In this video I walk through how I built easyflip.ai, an AI listing app for resellers, in 45 days with Claude Code and no coding experience. I recorded this in September 2025, when the app was basically ready for launch. It's for anyone who wants to build an app with AI, has never written code, and wants to see what the process really looks like, including the parts that went wrong.
Tools change fast, so check the current docs for anything I mention here before you follow along.
What I cover in this video
- 0:00 Where the idea for easyflip.ai came from and how Idea Browser helped me validate it
- 0:46 My reselling background and the listing problem the app solves
- 2:16 Idea Browser's build prompts and why the PRD matters most
- 3:47 Running
/initin Claude Code to create the Claude markdown file - 5:20 Where the app is after 45 days and what a hero function is
- 6:54 The Next.js starter kit that made the difference
- 8:26 The landing page, the waitlist, and why audience comes first
- 11:30 My debugging loop with a Claude project and Claude Code
Pick an idea you can distribute
The idea for easyflip.ai came from Idea Browser. It was first presented there as a decluttering app, and the trending keywords gave me some insight into whether it was worth building at all.
It also matched my background. I've been a reseller on eBay, Poshmark, and Mercari for 5 years, and we built a full-time business doing it. The most tedious part is creating listings. So I asked how we could use a vision model: you upload photos from your phone, and the AI generates the title and item specifics and prices the item.
The other reason I chose it is audience. Our reselling YouTube channel has 17,000 subscribers, I have an email list of over 4,000 people in the reselling niche, and we run a free community of 1,400 resellers. The whole point of the landing page is to collect emails, and the waitlist has passed 200 people without much marketing. Distribution is the most important part of building your idea. You don't want to spend all that time building and then have no way to sell it. If you don't have an audience yet, maybe you build one while you build the app.
Write the PRD before you build
Idea Browser has a build section with ready-made prompts for ad creatives, brand packaging, landing pages, and a PRD, the product requirements document. The PRD is the one I like most. It's the brain and the vision for your idea.
I pasted the PRD prompt into Claude and got back a long document covering product vision, target user, key differentiator, success definition, competitive advantage, market opportunity, and resource requirements. I would not have been that detailed on my own. Then I put the PRD into Claude Code and ran:
/init
That creates the Claude markdown file. Claude gets a lot of its information about how to build your product from these two files, so think of them as the road map.
This was my most important lesson. When we first started, we were building in Bolt (bolt.new), which was great, but we had no PRD and no specific information on how we wanted the product to turn out.
Focus on the hero function and start from a starter kit
The hero function is the main feature that makes your product valuable. For easyflip.ai that's the AI analysis: you upload photos, the AI analyzes them, and the listings go up on the marketplaces. Early on I wanted to build in so many little features that it broke the whole project. It took about 20 days to figure out that I only needed to focus on that one thing.
What made the biggest difference after that was the Next.js starter kit released by Rasmic. It lays out what an app needs to function, and it comes with those pieces already integrated:
- Authentication: Better Auth. Without authentication, users can't log in and you have no working app. From my understanding, Better Auth is great because you own your authentication.
- Subscriptions and billing: Polar for payments.
- AI: an OpenAI integration where you plug in your API key.
- UI: Tailwind and shadcn, so the site looks modern.
- Database and storage: Neon, Drizzle, and Cloudflare for image storage.
I had no idea what any of these were when I started. I also didn't know people shared plug-and-play repositories like this. If I were starting over, I would begin with this foundation and build on top of it.
My debugging loop with Claude and Claude Code
No one one-shots an app that provides value. Maybe the AI gets you 90% of the way there, and that last 10% is what makes the app make money. So you need good debugging skills, even if you're coming to vibe coding with no experience.
I set up a project in Claude for easyflip.ai and used it next to Claude Code. This loop was basically my whole life while building the app:
- When something breaks, ask Claude Code for the server-side logs.
- Copy the server-side logs and the console logs into the Claude project.
- Tell Claude what broke and ask for a prompt you can give to Claude Code.
- Paste that prompt into Claude Code and let it fix the issue.
The message I give Claude sounds like this:
Hey Claude, I'm giving you the server side logs. We had an error. This button broke. How can we fix it? I'm also giving you the console logs. Give me a prompt that I can give to Claude Code to fix this.
I use Wispr Flow for this, which makes it really easy.
Before these 45 days I didn't know what an API or a bash command was. I worked on this about 8 to 10 hours a day, hit a lot of errors, and even restarted in the middle of the project. That's how I picked up the skills: breaking things and learning from the mistakes. Next up is a beta test with about 5 to 10 testers to squash more bugs before launch. I'm not in any rush to release it before it's ready.
Key takeaways
- Check that you have a way to reach buyers before you build. Distribution matters more than the idea.
- Write a PRD first, put it into Claude Code, and run
/initso Claude has a road map. - Pick one hero function and get it working before adding small features.
- Start from a starter kit that already has authentication, payments, database, and storage set up.
- Learn to debug by sending server-side and console logs to Claude and asking for a prompt to give Claude Code. Expect frustration and keep going.
FAQ
Can you build an app with Claude Code if you have no coding experience?
Yes. I built easyflip.ai in 45 days without knowing what an API or a bash command was when I started. It took about 8 to 10 hours a day and a lot of errors, so you need a tolerance for frustration.
Why do you need a PRD before building with Claude Code?
The product requirements document is the brain and vision for your idea. Claude Code gets a lot of its direction from the PRD and the Claude markdown file that /init creates, and my first attempt in Bolt struggled because I didn't have one.
What is included in the Next.js starter kit used in the video?
The starter kit from Rasmic includes Better Auth for authentication, Polar for payments, an OpenAI integration, Tailwind and shadcn for the UI, and Neon, Drizzle, and Cloudflare for database and image storage.
How do you debug an app when you can't read the code?
I ask Claude Code for the server-side logs, paste those and the console logs into a Claude project, explain what broke, and ask Claude to write a prompt I can give back to Claude Code to fix it.
Transcript
Read the full transcript
0:00In this video, we're going to be going over my journey over the last 45 days from building this AI app from no coding experience to where we are at now with basically ready for launch. So, I want to walk you through the process of the initial concept of the idea to launch. So, the app is called easyflip.ai and this idea came from ideabrowser.com. If you have not heard about ideas.com, you need to go check it out. It is amazing for taking your idea, validating your idea, and taking the next step to making that come to pass. So, this really helped give me some insight as to if this idea is even worth building. The most important part of building your idea is distribution. If you have a
0:46great idea, but no one wants it, it's going to have a hard time being successful. So, the app idea idea browser sparked it. And you can see here there's trending keywords. This is a a needed thing. It was initially presented as like a decluttering app. And I have a background in reselling. I have been a reseller on marketplaces like eBay, Poshmark, and Merkari for 5 years. We've built a full-time business doing that. And the most tedious thing is creating listings. No one likes doing that. It's very old and outdated. So I was like, well, how can we use a vision model where users can upload photos from their phone and the entire process is handed off to the AI? It generates the titles, the item specifics, prices your item,
1:31and you set it and forget it. Wait for the item to sell. That's what we've built with easyflip.ai. And it took 45 days of painstaking, errors, bugs, restarting over even in the middle of the project. And that's why I'm making these videos, first of all, is to show people like yourself, maybe you have no experience vibe coding or any kind of development, and just showing that you can do this as well. The thing you're going to need the most, though, is a tolerance to frustration. Because holy moly, guys, this was very difficult, very frustrating, and it all just comes down to a skill issue. I lacked the necessary skills to put these ideas and concepts together for them to work. So before 45 days, I didn't even know what the heck an API was. I didn't
2:16even know what a bash command was. I had no idea. But hey, through trial and error, we've learned and now we know the basic concepts of how an app should be built. So this is amazing. Check out idea browser. Um there's more to it than just, you know, validating the idea. There is also um like a a building section where you can build your app out with prompts already made. So, ad creatives, brand packaging, landing pages. I think the most important one that I like the most is setting up the PRD, the product requirements document. This is like the brain and the vision for your idea. This is so important that if you miss this, it's going to be much harder for you to build a successful app
3:01because when you're using um you know something like cloud code or you're using something like cursor or you're using GPT5 to build out your idea, it's going to be have a hard time because you don't have a product requirements document. So this is the prompt that it gave me. I put that into Claude and then it gave me an amazing PRD. I'll just pull it up right here. Why are we even waiting? So the PRD after it was generated I put it into cloud code and you can see here this is basically it guys. This is the sauce for the sauce on the sauce right here. Look at this product vision target user key differentiator success definition strategic alignment users problem solved competitive advantage market opportunity resource requirements. I mean, this is a
3:47lengthy document that I definitely would have not been this um detailed and with things like Claude, you can build this out in 10 seconds. So, once you get the product requirement document set up into your Claude code, if that's what you're using, you're going to want to do a forward slashinit command and then it's going to initialize cla to be able to create a claude markdown file, which is also very important to have a very strong Claude markdown file. It's very similar to like the vision and the um brain of your product is um claude gets you know a lot of its information and how to build out your product out of these two files. So it's kind of like the road map, right? Um, so you can see there's a cloud markdown file built out.
4:34That was the most important learning lesson of building is that when we first started this project, we were building in bolt, bolt, new, which was great. But we lacked a product requirements document. We lacked specific information on how we wanted our product to turn out. So, I'm just showing you there because this is very, very important if you're wanting to build out your startup idea and you don't know where to start. I would highly recommend checking out Claude Code. Um, even if you have no experience. So, that's how we started it. Now, how we doing? Well, here we are, guys. Here we are. Let's just go to Easy Flip. This is where we're at right now after 45 days. Working about 8 to 10 hours every single day. Yes, we've gone camping. So, technically we took a break
5:20in there. Basically, every day we got this working. We actually just did the demo feature here. I just have a quick little demo demonstrating how the product works. It's really important that as a founder or a product developer that you focus on the hero function of your product. What is the hero function? What does that even mean? That means what is the main function or feature of your app idea that's going to make the difference. So that's why when we started this I wanted to build in so many little features that just broke the whole project and I wasn't focusing on the thing that actually made the product valuable which was the AI analysis. So people that upload photos the AI analysis happens and then it uploads
6:06them onto marketplaces. I just needed to focus on that for 45 days straight and that was it. And that's what we did after about 20 days of figuring that out. So now that we have that all dialed in, we have the main hero function of our product working. And it's amazing. Um, it really saves my time personally because I'm listing actively on eBay and Poshmark and Merkari. All I have to do is take photos and everything else is filled out. And so, um, you really got to focus on what is the what is the main part to your app that's going to make people want to purchase it, right? Um, so yeah, that's amazing. And of course, here's like the dashboard. Really simple. Um, I do want to get into what really made the difference in this app was this git
6:54this git repository that was released by Rasmike or aka Mickey on X. He is a great kind of teacher of just how this all works. Like what makes a good app? What does it need to have a good efficient app where things don't break? And here we have a Nex.js starter kit. This explains it all and brings so much clarity for people like me that have no idea what we're doing and just lays it out. Be like, okay, this is what an app needs to function. The first thing your app needs to function is authentication. If you have no authentication, users can't log in. Their password and information isn't stored. There's no way to verify if that's the person logging in. Therefore, you have no working app.
7:40Authentication, you use better. This starter kit, I'll link it down below, uses better off for authentication. And from my understanding, better off is great and versus other authentication platforms because you own your authentication. Um, so subscription and billing. It uses polar for payments, which is really cool. This is all integrated in the repository. It even has an open AI integration where you just plug in your um secret key, your API key and you can have all of those features. Also, modern UI UX is it uses Tailwind, Shad CN. Um so the website looks very modern, very updated database and storage. You got to set up your neon um drizzle and cloudfare for uh image
8:26storage. And understanding how that all works was key. And I had no idea what these were when we first started. And it just became more clear to me that this is what you need. This is like the foundation of your what you're trying to build. You got to have those three four things for it to work. And now that we have that all built, um, we have a nice looking landing page. Of course, we still need to make improvements, but I think, you know, starting off, this is good. The whole point of the landing page is actually just to collect emails. We've surpassed over 200 people on the wait list now without even really mentioning it, marketing it too much. I do want to say that audience is key when you're building an app and we only chose to build this was because we already have a YouTube subscriber base of 17,000
9:11subscribers on our reselling channel. I have a email list of over 4,000 emails of people that are in the reselling niche that would find this really helpful. And then we have a school community of one 1,400 people, a free community of resellers that would, you know, like to buy back some of their time and not spend so much time listing manually. So that's why we we initially were like, you know what, this just aligns with what we're currently doing. And I think that's so important when you're building anything out is that you need to make sure, do I even have an audience? You know, you don't want to spend so much time building something and then when you go to distribute it, you you have no marketing channel. you have no way to sell it, you know. So, I think that's they go hand in hand, you know. Maybe you're building while building the audience. Um, but that's
9:57why we gravitated so much to this idea because it just fit in and not only through the idea browser with it being validated. Um, it just made sense and it probably was a little bit ambitious to choose as my first project ever being built, but I did learn so much during this. And we're in the probably that beta testing phase here in the next week where we're going to have about 5 to 10 beta testers be able to use our app to be able to squash any more bugs and you know kill any more errors before we actually launch this thing because I want it to be as perfect as possible and just give people the best experience you know for what they pay for right I don't want to be releasing something that's not ready to be released. I'm not in any rush to release this by the way. So yeah
10:43that's just really cool walking you through the process. So download that Nex.js starter pack if you're trying to build something. Really cool, really easy to use. Maybe I'll make a video on on our next project that we start exactly how to set up that starter kit. Um, comment down below if that would be helpful. But that's how I would do all of this starting over would just be getting that foundation and building with that. So, and that's something I didn't know about. I didn't know that people had, you know, repositories that were like plugandplay ready to go. So you at least have everything, you know, your authentication, your database. Um, so that's already built out and that was really cool. So another helpful thing that was detrimental to building this out was just working with Claude and
11:30setting up a project within Claude to help guide me along this project. So I have an easy flip. I claude project that just allows me to just talk with the data you know so as I'm building out in cloud code anytime I would get an error let's just say you know an error with the API or whatever I would go ahead and highlight you know an error that they saw on the server side logs I would tell claude hey claude code hey give me server side logs because there's an error I would copy paste the server side logs I would then go into claude and you know tell Claude, "Hey Claude, I'm giving you the server side logs. We had an error. This button broke. How can we fix it? I'm also giving you the console
12:16logs. Give me a prompt that I can give to Cloud Code to fix this." Boom. Using Whisper Flow. It makes it really easy. And then, you know, just copy paste the console log or whatever in there. And then you hit enter. And that was basically my whole life by building this app was just understanding the process on how apps are built and using Claude with Claude code to help me figure out what was going on because you're not going to know it all, especially if you are starting out with no experience, but using Claude to help you understand some of the concepts of what why things were breaking, was it me breaking them or you know, things like that. And I would say what I've learned from all that is you really need to have great debugging skills
13:02because you're going to come across errors and you're going to come across problems. No one oneshots an app provides value. You know, if you're going to oneshot an app and it gets it to a point, maybe it gets it 90% of the way there, that extra 10% is what makes the app really print money, right? So, um, we got our app 90% of the way there, but you need to understand like what are some of the reasons why it's not working. So using cloud in conjunction with cloud code to help you figure out why it's not working, you know, debugging and logging is going to help you know solve these issues. So checking the server side logs, checking console logs, copying pasting those in one uh prompt to give to Claude and then I would tell Claude, "Hey, give me a prompt I could give to Claude code to go
13:47fix this because I just it's a skill issue, right? When you're building anything with no experience, you don't have any skills. and how do you get skills? For me, it's breaking things. It's making errors. It's learning from the mistakes. And by doing that process, it helped me understand a lot of the ideas and concepts, the principles of how to build a successful app and what all goes into it. So, use these AI tools to your benefit, to your advantage, especially as you're growing and building an app. I'm going to be making more videos, you know, every week just detailing the process and showing you. I already have a few side projects already picked out as we're basically getting ready to launch this thing. It's kind of a sad thing because I've been so excited to share with you. Maybe I'll make update videos in the future of how it's
14:32going, our MR, how many signups we have, and and then we're going to be starting and building out our next app idea, which I already have a few picked out that I'm super excited. And they should be pretty relatively easy to build. And I'm going to be sharing the journey all along here. I want to know your guys's thoughts. Where are you at in your journey? How are you building things? Maybe some suggestions and tips for me because I'm still learning. But I'm so excited to keep jumping into these videos and building alongside you guys. I hope you have a blessed rest of the day and we'll see you in the next
Ready to start building with AI?
Join the Shipping Skool community free. Paid members get live bootcamp calls every week and weekly-updated courses.
Join Shipping SkoolKeep reading
How to Start a SaaS With AI: Idea Browser, PRD, and Claude Code
Validate an idea first (I use Idea Browser), turn it into a PRD with Claude, then build in Claude Code on top of Ross Mike's Next.js starter kit. Hook it up to GitHub, a database like Neon or Supabase, Vercel for hosting, and Better Auth for login.
How to Build a SaaS MVP with Google AI Studio and Idea Browser
I use Idea Browser to pick a validated idea (an AI marketing tool for people selling their home without an agent) and grab its PRD and build-kit prompts. Then I paste them into Google AI Studio to build the landing page, Firebase login, dashboard, AI listing wizard, and leads page in under an hour.
Next.js Starter Kit With Claude Code: 3 Apps Toward $1k MRR
I used Michael Shimeles' Next.js starter kit, which comes with authentication, payments, and a database already built in, to ship three apps in 90 days. In this video I share where each app stands and how I plan to market them toward $1,000 in monthly recurring revenue.
