I Built an AI Web App in 3 Weeks For $322
I just launched ReelPal AI, an AI video generator and editor. Here's how I built it, the tools used, and my plans to grow the app.
Hey friends, I just launched a project I have been grinding on for the past three weeks. It is called ReelPal AI. The platform lets you create short AI videos that come with voiceovers (feed your script), background music, and are powered by the best video models available right now.
Oh, another AI video generator? Aren’t there already too many?
I know. But ReelPal isn’t just a simple AI video generator. Users can add voice narration, background music, and use a bunch of high-quality templates based on the most trending AI videos online. This makes it super easy to build product ads, mini films, music videos, or trending clips in a few minutes.
The idea of ReelPal started when Google released Flow. It is a browser-based video editor and AI video generator powered by Veo 3. It looked great. A tool that can stitch AI videos together and export them in one output is exactly what I wanted. But there was one problem. It only supported Veo 3. That model is expensive. You cannot switch to more affordable ones like Kling 2.1 or Hailuo 02. It also had no voiceover or background music options.
So I decided to build my own.
In this article, I will talk about how the idea came together, how I built ReelPal in just three weeks, how it works, and how you can build something similar.
Disclaimer: I am not affiliated with or sponsored by any of the platforms mentioned in this article.
Let’s get into it.
ReelPal AI is Based On An Open-Source Project
Thanks to the open-source video-starter-kit project shared by the popular AI API service provider Fal AI, I was able to build and launch ReelPal in two weeks.
Here’s what the original project looks like:
Here are the built-in features of the base project:
Browser-Native Video Processing: Seamless video handling and composition in the browser
AI Model Integration: Direct access to state-of-the-art video models through Fal AI (Minimax, Hunyuan, LTX video)
Advanced Media Capabilities: Multi-clip video composition
Audio track integration
Voiceover support
This is what ReelPal’s video editor page looks like:
It looks similar to the original, but I’ve actually made a bunch of changes, especially on the backend:
I switched from a local database to a server database, which allows a user to save a project and load it on another device.
I tweaked the user interface, made the components more rounded, made text larger, and added a bit of animation to improve the look and feel.
I introduced more powerful video models such as Veo 3, Kling 2.1 Master, and Hailuo 02.
Introduce authentication so users can sign up and sign in.
Allow loading projects from templates.
The video-starter-kit project is under the MIT license, which gives me the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software.
If you intend to create a similar app or simply learn about web development, I encourage you to go download or fork the source code while it is still available.
Let’s Talk About The Tech Stack
Here is a list of all the frameworks and platforms used in the ReelPal project.
NextJS React Framework
TailwindCSS for styling
PostgreSQL + Neon DB
Fal AI + OpenAI for AI backend
AWS S3 for storage
Authentication through Neon Stack
Hosted on Vercel
For coding, I use the Cursor IDE to build everything. I could have done it with Lovable or Bolt, but Cursor is currently the best tool for vibe coding. You get full access to the source, so you have more control, and you can switch between large language models easily. I mainly use Claude 4 Sonnet in Thinking mode, but sometimes I switch to Gemini 2.5 Pro when Claude cannot fix a bug or when Anthropic is down.
For streamlined development and deployment, I use GitHub with Vercel. Both work well together, so version control and code management are smooth. GitHub also includes Copilot AI, which makes documentation painless.
For authentication and the database, I now use Neon because it is bundled with Vercel Pro and is easy to set up. I no longer open the clunky Prisma Studio. Neon’s browser based studio is pleasant, and the built in authentication service makes user management simple.
For storage, I stay with Amazon S3 mainly for cost. I tried Vercel Blob and Supabase Storage, and they burned through funds fast. With S3 buckets, I pay less than ten dollars a month even with hundreds of images and videos stored.
To summarize the list of tools I mentioned above, here is a simple table:
Anyway, the tech stack is up to the developer. I hope the details I shared above help anyone get started with building AI-powered web applications.
Key Features of ReelPal AI
ReelPal provides all the tools you need to create, edit, and share cool AI videos that drive traffic to your website.
Here are the key features:
Best Video Models: ReelPal is powered by the latest video models like Veo 3, Kling 2.1, Hunyuan, and more.
Best Image Models (coming soon): Choose from the most powerful image models like Imagen, Flux, GPT-Image-1, and more.
Add Voiceovers: Want an AI-powered narration? Just feed a script and get high-quality voiceovers in seconds.
Add Music and Sound Effects: Generate AI music and sound effects to make music videos, short films, or ASMR clips.
Edit and Export Videos: Export your stitched videos and overlayed music or voiceovers to an HD video file.
Share to TikTok, Reels, and Shorts (coming soon): Share directly to social platforms, schedule posts, or download for later use.
These are only the basic features on launch. Check out the “What’s Next For ReelPal AI?” section of this article to know more about the upcoming features I am working on.
How To Generate Videos on ReelPal AI
To get started, head over to ReelPal and create an account for free. Once you’re on the homepage, click on the “Create Video” button, and you will be redirected to the video editor page.
Next, click on the Generate button, and a Generate Media panel on the right side will appear. In the Video tab, set the video model to Veo 3 and add the text prompt or description of the video you want to create. Make sure to enable the embedded audio by enabling the Generate Audio toggle.
Next, describe the video you want to generate. Here’s an example:
Shot in extreme macro perspective, a flawless, crystal-clear, and detail-rich glass strawberry rests on a wooden cutting board bathed in warm lighting.
The camera gently pushes and pulls, capturing the rainbow-like glimmers and prismatic refractions on the surface of the glass strawberry.
The angle alternates between top-down and a 45-degree side view, freezing the moment the knife slices into the fragile “glass strawberry,” revealing its intricate internal textures.
Each slow, deliberate cut produces a crisp “ting” — the sound of glass shattering — followed by the soft clinking of tiny shards and the deep, resonant scrape of steel against wood.
These sounds layer together to create a rich ASMR soundscape in an otherwise silent room. The overall pacing is slow and soothing, crafting a tranquil ASMR experience.
I designed ReelPal to offer as much control to the user. You can expand the Advanced Settings menu and modify the parameters of the video model, like the duration, aspect ratio, and negative prompt.
The advanced menu depends on whatever customization is offered by the video model. Some offers more options like changing the resolution, camera movement, and CFG scale.
Finally, click on the Generate button and wait for the final video to be fully rendered. Once done, you can click to play the video or drag and drop it into the timeline to preview the result.
Here’s what the final 8-second video looks like:
I love how the video turned out. The physics seems accurate, and the sound effects, while the knife cuts through the glass watermelon, are weirdly satisfying. No wonder these kinds of videos are going viral on social media.
I also created the Templates page so users don’t have to start from scratch. They can explore the list and choose which ones they want to recreate.
To import the project into your own video editor, click on the “Use this template” button. You should be redirected to the video editor page.
Once the project is loaded to the video editor page, you can do whatever you want with it. My advice is not to export them immediately. Try to re-generate the video and audio clips so they are unique.
In a matter of minutes, you now have your own version of the most trending AI videos on the internet. No need to worry about which prompt to use. You also save time from stitching a bunch of AI videos into one.
The Right Way To Go Viral on TikTok
If you want to use ReelPal to make viral videos, you may wonder whether to buy or create TikTok accounts. It is always better to create new ones. A fresh account starts with a clean reputation. If you buy an account, it could be flagged, throttled, or shadow banned. That is harder to fix than starting new.
Warm the account before posting: scroll your For You Page, engage naturally, and avoid uploading anything for the first few days. This is very important to show TikTok that you are a real human using the platform. Don’t overthink it.
After that, begin posting one or two clips daily and keep interacting for at least five minutes per session.
ReelPal AI Pricing
Now let’s talk about pricing. ReelPal has three simple plans depending on how much content you want to create.
The Starter plan is priced at $19.99 per month. It gives you 200 credits and full access to all the premium AI video models on the platform. This plan is good if you’re just getting started or only planning to create a few videos here and there.
The Growth plan costs $49.99 per month and gives you 500 credits. This is the best option for content creators. On top of the same premium model access, you also get the ability to publish directly to TikTok, which is great if you’re making short-form content often.
The Pro plan is $99.99 per month and gives you 1,500 credits. It includes everything in the Growth plan, plus support for scheduled posting. This is perfect for teams or agencies that manage a lot of content and want to automate their posting schedule.
All plans give you access to the same high-quality video models. The only difference is how many credits you get and the posting features available. You can cancel anytime or switch plans based on how much you’re using the app.
Cost of Building and Maintaining a Web App
You might be wondering how much it really costs to build a web application and then keep it alive after launch. The honest answer is that the budget depends on the scope, the number of moving parts, and the kind of user experience you are aiming for. There is no single price tag that applies to every project.
A big project with a team of engineers could easily burn $10K. But a smaller project, like ReelPal AI, with a single developer cost roughly ~$350 in the first three weeks.
Let me break the numbers down so you can see exactly where the cash went:
Cursor monthly subscription: $20
Vercel Pro monthly subscription: $20
Fal AI credits for testing the AI models: $100
Domain name from GoDaddy: $180
AWS S3 storage: $2.44
The total expense is $322.50
That’s relatively cheap. Hiring an experienced software engineer to do the same work would easily push the cost past five thousand dollars.
Maintaining the web app is not expensive either. Beyond the fixed subscriptions, I only spend on adding new templates and on API fees, which rise naturally with paying users.
What’s Next For ReelPal AI?
First on the roadmap is social posting. I have already filed the TikTok integration, and the request is sitting in their review queue. As soon as it comes through, users will be able to push their AI videos straight from ReelPal to their TikTok feed without extra downloads or workarounds.
After that, I am adding a posting calendar. I am building a clean calendar view where every square is a slot on your content timeline. You will drop a finished video onto a date, pick the hour, and ReelPal will publish it while you sleep. This would be very helpful to content creators and marketing people.
I also plan to add support for image generation. Many video models work better when they start with a still image. Making that image inside ReelPal keeps everything in one library and lets you try image-to-video prompts.
A text-overlay tool is also on the way. This is useful if you’re working on product video ads, short films, or any promotional content that shows text.
I will also be adding more pages to the web app, like documentation, blog, guides, and more. These are all important, but I don’t have all the time in the world, so they have to be pushed back to a later release.
Finally, I am doubling down on marketing. Code is only half the job. I will share progress threads, demo clips, and honest notes on X, Medium, and everywhere I hang out online. If you try the app and leave feedback, you steer the next sprint. If you subscribe, you fund a few more late-night coding sessions, and I appreciate every single one.
Final Thoughts
Building ReelPal is such an incredible journey. It was a super busy three weeks, but it was worth it. I love how the project turned out and hope many people will enjoy using it.
ReelPal serves three clear goals for me.
First, it adds real weight to my portfolio, joining the growing list of vibe coded web apps I have shipped. Anyone who asks what I can build now has another working example.
Second, I expect the paid plans to provide real income. As a solopreneur I need each project to stand on its own while generating income, and ReelPal is set up to do that.
Third, the process taught me new skills. I worked with new AI video models, new tools, and discovered new workflows to build apps faster and more efficiently. Those lessons will help be build better apps in the future.
In the next couple of days, I will be doing a bunch of new feature implementations, bug fixing, UI improvements, and lots of marketing. I intend to use Medium and X as my primary platforms to spread the word. I am also planning to create TikTok and YouTube short videos, hopefully, bringing more eyes to the website through viral videos.
I share the journey because the AI community grows when builders talk openly.
I spent the past three years building in public. The AI community is comprised of developers and founders sharing their stories, strategies, and even their failures.
Building in public is how I scale my products and grow my network. If you are reading this and interested in building cool projects with AI, I encourage you to be part of the AI community on X, Reddit, and Medium.
If you are interested in building a similar project, grab Fal’s open-source video-starter-kit project and start tweaking it to be more aligned with your ideas. Again, you don’t need to be an experienced developer to do this. You just have to be aware of the right AI tools to use and how to use them.
If you have any questions, please let me know in the comments.
Hi there! Thanks for making it to the end of this post! My name is Jim, and I’m an AI enthusiast passionate about exploring the latest news, guides, and insights in the world of generative AI. If you’ve enjoyed this content and would like to support my work, consider becoming a paid subscriber. Your support means a lot!