辞职以发展他的五位数 MRR 产品组合

Building a portfolio

I'm a self-taught dev based in Cyprus, building a few AI SaaS products at once.

The main one right now is Motionvid.ai, an AI video and motion graphics tool. We ran a highly successful AppSumo campaign that brought in $120k. And we've grown 40% month-over-month since then, now exceeding 100k signups and $7k MRR.

I also run Sally POS, a point-of-sale system for restaurants and bars, built with React Native. It operates in over 400 locations across South Africa, Cyprus, and Australia, processing daily transactions. It's bringing in a 5-figure MRR.

And Framekit.ai is an AI website builder for creatives, powering over 200 live sites. It's bringing in $1k MRR.

Taking the leap

I was working as a software engineer at a forex company, earning a solid salary when I started Motionvid.

A friend was building a course and wanted an easier way to make motion graphics without touching After Effects, so I vibe coded a test version with Sonnet 3.7 and Cursor in March 2025 to see if it would work. He made a YouTube video about it, which sent us our first real traffic. The name was also an exact SEO match, so search also fed traffic to a high-converting landing page. From there, it kept growing organically.

Once I saw the SaaS taking off, I knew where to focus my time. I quit, took the pay cut, and went all in on Motionvid.

While that happened, I also built Framekit on the side. And Sally POS is the oldest of the three. I started it six years ago as my first business, and it's still running today.

Building the product

Building the initial version of Motionvid took about three weeks. I was vibe coding it with Sonnet 3.7 in Cursor, so the only real cost was a Cursor subscription and about $300 for tokens. No team, no outside help, just me testing whether the idea worked.

AI made a lot of mistakes at the time and produced bad (or ugly) code, so I continually corrected it. Now AI is much better at it.

Since that first version, the stack has grown to include Next.js and React for the frontend, Golang for backend services, Firebase/Firestore for data, Cloud Run for deployment, Stripe for billing, and Cloudflare in front of everything.

I built our own MCP integration so we can call video generation programmatically instead of just through the UI. Model routing has been the trickiest part; we mix a few models depending on the task. Getting image generation to work well required testing several vision-capable models before we landed on what we use now.

Our biggest technical challenges so far have been cost accuracy and defect hunting as usage scaled. There was a defect where scenes double-committed on longer video jobs. And we caught a bug where credit cost estimates were significantly overinflated compared to actual usage. If I hadn't caught it, it would've quietly eaten into our margin unnoticed. Track your costs. Every model call, every agent step, every workflow you automate. It adds up fast and remains invisible until you look.

We fixed both, but these issues only appear once real users push significant volume through the system.

Subscription businesses

Motionvid charges monthly for video and motion graphics generation, with tiers based on usage/credits. It started free while I tested the idea, then became paid once YouTube video traffic and SEO showed people wanted it. Growth has since come from organic search, affiliates, and the AppSumo campaign. The AppSumo campaign brought in a lump of lifetime-deal revenue, which we're reinvesting into UGC to keep the organic engine going. Revenue expands here through usage-based tiers; as customers generate more videos, they move up, so the model scales with the value they receive.

Framekit is also subscription-based, but it has been the slowest of the three to grow. Customer acquisition, rather than the product itself, has been the bottleneck. It remains active with minimal ongoing development rather than active investment.

Sally POS is different; it's B2B software sold to hospitality venues. Revenue comes from recurring subscription fees per location, not per user. Expansion means adding more venues and countries rather than upselling existing customers on usage.

Social media, SEO, LTDs, and affiliates

Motionvid grew through two main tactics. A friend made a YouTube video about the tool right after I built the MVP, which generated our first wave of users and proved the idea's viability. From there, SEO became the main channel. The product name exactly matches what people search for, so we started ranking without much deliberate SEO work; the name did the heavy lifting. A high-converting landing page consistently turned that search traffic into sign-ups.

Once we had traction, we ran an AppSumo campaign. This brought a large batch of new users and provided significant capital to reinvest, generating six figures in lifetime revenue from that alone. We are now investing that money into UGC content to extend the same organic playbook that already worked, rather than switching to paid acquisition. Affiliates also naturally picked us up, finding us through the same YouTube and search presence rather than through any formal affiliate program.

Framekit and Sally POS experienced different growth paths. Framekit's growth has been slow; customer acquisition, not the product, is the bottleneck, so it has not had a strong channel like Motionvid. Sally POS grew through old-fashioned direct sales into hospitality venues across South Africa, Cyprus, and Australia, expanding location by location rather than through any online channel.

My advice: Do not force distribution before you prove the product is worth distributing. Motionvid grew because a real person made a video about something that worked. That is a stronger channel than any paid campaign I could have run at that stage.

Unfair advantages

A few things have been genuinely helpful — some from skill, some from luck.

Vibe coding with AI tools proved the single biggest unlock. Going from idea to working MVP in three weeks, solo, with just Cursor and a few hundred dollars in tokens, allowed me to test ideas without needing a team or real capital first. That changed the entire risk profile of starting something.

The exact-match product name for Motionvid proved a lucky decision, becoming a real advantage. SEO started working almost passively because people search for the name itself. I didn't plan that as a growth strategy, but it's been one of the strongest channels since.

A filmmaker friend, genuinely excited enough about the idea to make a YouTube video about it, was a significant force outside my control. That video brought in the first real users. No amount of my own hustling would have replicated that kind of organic, third-party trust.

On the skills side, building across a few different stacks (React, React Native, Go, Firebase) let me move fast on whichever product needed attention without waiting on other people. And being monetization-first from day one — thinking about how something makes money before I get deep into building it — has saved me from sinking time into things that would never work as a business.

Five pieces of advice

Vibe code your MVP first; don't overthink the stack.

I built Motionvid's first version in about three weeks with Cursor and roughly $300 in tokens. You don't need a team or real funding to find out if an idea has legs; you need a working thing fast enough to let real people react to it.

Track your token and AI agent costs from day one.

This isn't obvious until it bites you. We had a cost estimation bug that wildly overinflated costs compared to actual usage. If we had underestimated them, it could have quietly wrecked margins before anyone noticed. Set this up early, not after you scale.

Don't force distribution before the product earns it.

Motionvid's first real growth came from a friend making a YouTube video because he was genuinely excited about it, not from a marketing plan. If the thing you built is useful, word tends to travel faster and cheaper than anything you'd pay for.

Be monetization-first, even as an indie hacker.

Know how something makes money before you go deep on building it. It's saved me from wasting months on ideas that were never going to be a real business, no matter how fun they were to build.

And don't be afraid to run a few products in parallel.

Not every idea needs to be your one big bet. Some things, like Framekit, grow slowly, and that's fine; they can keep running with minimal upkeep while your main focus goes wherever the traction is.

What's next?

My goal is to hit $100k MRR for Motionvid.ai and then find a strategic acquisition. But I really enjoy building it, so I'm not sure if I want to exit.

You can follow along on my personal website, Twitter, or Instagram.