Building a product with AI in four days and hitting $12k in four weeks

Sebastian Volkis built a product in four days using AI. Four weeks later, TrendFeed, has made $12k. And his AI coding course, CodeSpring, has made $45k since January.

Here's Sebastian on how he's doing it. 👇

Getting started

For as long as I can remember, I’ve been involved in random business ventures. I started off selling digital products, ran a YouTube channel where I made photography tutorials, and later, I ran dropshipping stores.

After leaving university, where I studied physics and astrophysics, I started working as a freelance media buyer. I absolutely hated it. So, I quit and spent six months unemployed, trying to figure out how to build apps.

I came across various AI coding tools and no-code platforms like Bubble, which is where I built my first app, ChatIQ.ai. In the first two months after launching, I reached around £7.5k ($9.7k) MRR.

My biggest challenge was that I couldn’t build features as quickly as my competitors. I was completely new to this and had no technical experience. Soon, customers started churning from ChatIQ and switching to competitors.

To try and keep up, I began using GPT-4 to write APIs in Python and create feature extensions that could integrate with my Bubble app. These worked, but they were slow and very buggy.

Eventually, I realized I needed to build apps using full-stack code. The problem was that I didn’t want to spend months learning to code from scratch.

Learning to build with AI

Because of what I’d managed to do with ChatIQ, I figured I could use AI to code for me. So, I learned how to build apps with AI and Next.js, just by building lots of micro-SaaS products to get the reps in. I'm at a point where I know how to get AI to write secure, efficient code that can run in a production environment.

As I was learning this, my twin brother, Matthew, came to me and said, “Let’s team up. I’ll handle the marketing, you build the product.”

Our goal: Build and exit a SaaS in under 90 days.

$12k in 4 weeks

So we started working on TrendFeed.app, a tool that aggregates viral, trending news articles and programmatically generates short-form content from those sources.

For beginners, the most effective way to get traction is by leveraging viral, trending news to generate views. That’s exactly what TrendFeed was designed for. It aggregates news articles, runs them through an algorithm to assess virality, and gives users the tools to automate content creation. We knew that if we could help users start making videos quickly and see results straight away, the tool would take off. So TrendFeed was built to make content creation easier and more accessible.

I didn’t want TrendFeed to be just another generic AI video creation tool. There are hundreds out there, and most don’t offer anything unique. Instead, we focused on analyzing credible sources, using AI to identify viral trends, compiling visual assets from these stories, and turning them into high-quality videos that people on Instagram and TikTok actually want to watch.

We gave ourselves a week to launch. I managed to build an MVP entirely with AI-generated code in just four days. Matthew spent the week driving traffic to a waitlist for our launch event through his Instagram, an email list, and new TrendFeed accounts on Instagram and TikTok, while I pushed some traffic from my own Instagram.

We launched on February 27th with a webinar, selling lifetime deals for TrendFeed, and managed to generate £5.5k on launch day.

Since then, we’ve been growing TrendFeed’s socials so we can market through those channels instead of relying on our personal brands. I’ve overhauled the product, added a lot of new features, and we’ve just started charging for yearly subscriptions.

In our first four weeks, we’ve made £9,222 ($12k) in revenue.

Not just another "vibe-coded" app

I wanted to make sure this wasn’t just another basic "vibe-coded" app that would break in production. Throughout the build, I focused on implementing security measures and guiding the AI models to ensure the product was structured properly. None of the code is handwritten, but I still needed a clear understanding of how to structure the app to make it all work.

There’s still a lot to do, but our early adopters have been patient and provided incredibly useful feedback.

When I started building apps, I was using Bubble. It was great for getting to grips with SaaS and launching products quickly, but it was slow, buggy, and you don’t own the code.

TrendFeed is fully coded using AI. It’s built with Next.js, React (using ShadCN for UI), Supabase for the database, and Vercel for deployment. Our stack is starting to evolve as we integrate Clerk for user authentication and Whop for payments.

Since this app is built entirely with AI-generated code, I know there are developers out there who enjoy poking around and hacking "vibe-coded" apps. I fully expect someone to try that with TrendFeed at some point, which is why I chose Clerk for its strong security features. It's also the reason I spend around twenty to forty percent of my time researching and implementing security into TrendFeed. Luckily, I already had a solid foundation in this area from previous projects.

Launch to a following

Our launch was actually very simple. We knew we’d sell through a live sales event (webinar), and all we needed were leads.

Three years ago, when I launched ChatIQ, I used TikTok, posting short video ads that showed what the product did. This let me tap into the massive ChatGPT trend and bootstrap the app to £7.5k MRR with £0 spent on marketing. So, I knew that if I ever launched something new, I’d need an audience. And I spent two years focused entirely on that.

I built a following across Instagram, TikTok, and YouTube. Today, I have 50k, 30k, and 4k followers on those platforms, respectively. Those were the sources of leads for TrendFeed's first launch.

That said, we want to eventually exit from TrendFeed, so everything we’re doing is designed to get the best possible multiplier when we sell. And that means we can’t be scaling this SaaS through our personal brands. So after launch, we shifted all our marketing to new TrendFeed social accounts.

As a result, revenue took a big hit. This drop in revenue is directly tied to a drop in views. There’s just no way an account with 100 followers can compete with one that has 50,000, at least not in the first few weeks. But this will scale as the accounts grow. Our sales funnel is already dialed in. All we need now is more traffic. Our plan is to scale this through influencers, hiring creators, and eventually, running ads.

I see a lot of people jump into short-form content thinking it’s a magic bullet for marketing. And while that can be true — it was with my first SaaS, ChatIQ — it only works if you’re tying trends directly to your product. Otherwise, it takes a lot more effort.

Consistency, while cliché, is the only strategy guaranteed to lead to long-term growth.

Use a webinar funnel

Our launch revenue wouldn’t have been possible without a webinar funnel. We’ve been creating webinars for years and have refined the process into a bit of an art form. But it all started with reading Expert Secrets by Russell Brunson.

That said, webinars are hard. In the past 12 months alone, for my business outside of TrendFeed, I’ve created over 35 different versions of webinars. Each one is around 300 slides, so it’s a serious amount of work. For most beginners, though, they could get similar results just by running weekly live demos of their app, signing up leads, and walking them through the problem and how their SaaS solves it.

Having just 20 leads on a live demo can massively increase conversion rates. It gives you the chance to handle objections in real time and build trust with potential customers in a live setting.

Since we want to control the sales process through webinars, we're in a closed beta. There’s no way for people to browse the site or try out the tool directly.

This is intentional, we want to control the sales process through webinars, as it gives us the best chance of converting leads into paying users.

It’s not the most traditional approach, and because the website is focused on collecting leads for our free training event, a lot of people at first didn’t believe TrendFeed was a real product. Some thought we were just selling a course. Moving forward, we’re planning to release a free plan that gives users five credits to test the app.

Don't lose control of the sale

Let's talk more about controlling the sale.

Our launch sold 22 customers on a £249 lifetime deal for TrendFeed. That generated over £5,500 in revenue from just 250 leads and 400 clicks to our site. Now, let’s compare that with a traditional SaaS funnel. With the same 400 clicks, you might convert five percent into free users, and ten percent of those into paying customers. That gives you 20 free users, and just 2 paying customers at £20/month, or £40/month in revenue.

So, in summary:

• Controlled sales funnel: 400 clicks = £5,500 in revenue.

• Traditional SaaS funnel: 400 clicks = £40/month in revenue.
  To match our launch revenue, you’d need 52,000 clicks to the site.

When you’re launching, getting traffic is hard. So don’t send that traffic to a passive landing page with a payment link — you’re losing control of the sale. That kind of setup can work brilliantly at scale, but in the early stages, every lead matters. Every person who visits your site is a potential customer. Worst case, if they don’t like your product, just refund them.

Now let’s take it a step further. The initial launch came from our personal brands, so we obviously had an advantage. Our audiences know and trust us, so they’re more likely to buy.

Or are they? These sales events are built to convert cold traffic. So we wanted to put that to the test. Right now, all our marketing comes from Instagram and TikTok accounts that never show our faces. It’s marketing designed to attract the right audience, while separating ourselves from the brand to avoid key-person risk.

And guess what? This new traffic still converts at the same rate as our launch. These accounts didn’t even exist four weeks ago. Just by being consistent and posting valuable content, you can grow quickly. But only if you control the sales process.

Now, imagine what happens when you do get 52,000 clicks to your site.

Sell the offer

Lastly, make sure your content sells your offer, not your product. Your videos shouldn’t be ads for your SaaS. Think about how many times you scroll past or skip ads, and how frustrating it is when one shows up in your feed. If your content looks like an ad, people will scroll right past it.

Instead, entertain. Add controversy. Hook people with a problem, then sell them on a lead magnet.

Here’s an example of what not to do: Show a video demoing how to use your app or what your app does.

Here’s what you should do: Show the frustration your app solves. Make the viewer relate to it. Then offer a lead magnet. For example, a free class to learn how to solve that frustration. Push those leads into your live demo or sales event, and close the sale there.

Focus on distribution

One thing I see most beginners do is launch their products into the void and overengineer everything.

Your product doesn’t need to be amazing at launch. Just because you’ve built something great doesn’t mean it’ll instantly go viral on X. For every viral launch post you see, there are 10,000 other failed launches that same day.

Focus on distribution. Build a sales funnel that converts a high percentage of leads into customers. If you’re not able to scale views on your content, focus on improving your conversion rates.

Seventy percent of your time should be spent on marketing and sales. There’s no point building features for users who don’t exist.

And most importantly, give it time. We managed to scale TrendFeed to £9k in its first month, but that came off the back of ten years of learning and honing our skills. Don’t forget the work that happens behind the scenes.

What's next?

We want to get TrendFeed to the point where it’s consistently adding £20k in ARR every single month. Our initial goal was to exit after 90 days and become one of the first SaaS companies built entirely with AI and sold for six figures in under 90 days. But if we’re scaling well, we might hold onto it for longer.

Looking ahead, I’ll always be building SaaS tools, but I’m especially focused on AI coding. I know there’s a lot of pushback when it comes to AI-coded apps, but I genuinely can’t see a future where app development isn’t fully democratised by AI.

I’ll continue sharing my journey as I learn full-stack development while using AI to build, and I’ll probably start experimenting with agents soon to automate parts of my workflow.

You can follow along on Instagram, X, YouTube, and TikTok. And here's my cofounder's Instagram. You can also check out Trendfeed, ChatIQ, and CodeSpring, where you can learn to code with AI.