From Physics to Side Hustle: How Two Dads Built and Grew a Micro-Learning App

Meet Philip Sergelius.

Philip is 37, born in Finland, but living in northern Germany since 1992.

Philip and his co-founder Max have similar lives: two kids each, a background in academic physics, and full-time jobs in the semiconductor industry.

Their side hustle NerdSip gets built in whatever time is left over, which means early mornings, evenings after bedtime, and occasional quiet hours on a weekend.

The idea for NerdSip came from our own commutes. We were both spending half an hour a day scrolling feeds that left us with nothing, and we wanted that time back.

NerdSip is a micro-learning app that turns those minutes into general knowledge you actually keep, with lessons short enough to finish on the train and fun enough that you don't feel like you're studying.

If you google "best micro learning apps", you will get the same names every time: Duolingo for languages, Blinkist and Deepstash for book summaries.

Everyone else picked a lane. We used AI to remove the lane entirely.

Type in anything you're curious about, whether that's Roman aqueducts, how interest rates work, or the biology of sleep, and NerdSip generates a complete course for you: text, illustrations, quizzes, infographics.

On top of that sits the game layer, with streaks, rankings, friends, and item drops. I spent two weeks building a 3D character in Three-js with items, sets to equip as if it was Diablo. Nobody uses it, but its cool. And it was a good exercise. I think every app needs this extremely overdesigned feature nobody uses.

The part I get genuinely excited about is the algorithm, which Max built.

Every course you generate flows into a shared feed, so whatever you just learned gets passed on to other curious people, and in return your feed fills up with the courses others found most fascinating. The best ones rise to the top on their own.

Thousands of strangers are effectively building one giant pool of general knowledge together, and it grows every single day. Nobody curates it. It curates itself.

Almost everything has been organic so far, but we are looking into starting scaling with ads.

We've spent less than $500 on ads in total, spread across small tests on Reddit, TikTok, and the App Store, and none of them beat what we get for free.

One of the smartest things we did early was to build our own analytics dashboard (screenshot below). It tells us exactly where users come from, and the answer surprised us.

SEO and AI Search

SEO drives over 40% of new users, all flowing through our blog at nerdsip.com/blog, which has become the backbone of the whole operation. People land there searching for things like "how to improve my general knowledge" or "how to never run out of things to say", and enough of them stick around to try the app.

AI search is second at 28%, meaning ChatGPT and Google's AI answers citing us. Everything else is noise.

TikTok

TikTok is the strange one.

Our videos have passed 10 million views across four accounts;

Conversion is weak though. Roughly one new account per 10,000 views. We kept at it anyway because the format works and the views are free, and lately the curve has started to bend.

Recently we had our first day with 60 new signups from TikTok alone.

We’ve had some success with panoramic content. See the screenshot below or check it out on TikTok.

This specific slideshow scored some 170k views.

With the panoramic content format we regularly break 100k views. We just present facts, cool stories, and a CTA on the last slide.

We are building a brand on TikTok. Gaining followers for authority signals.

Also, after 1000 followers your link in bio becomes clickable 👍

YouTube

We also publish long-form YouTube videos (example below), generated programmatically with Remotion, and they perform surprisingly well.

Backlinks

We've spent a few hundred dollars on premium directories purely for the domain rating (DR).

My favorite ones are TinyLaunch and VerifiedTools, offering the best DR for the price. We also listed on eg. PeerPush and FoundrList.

A backlink keeps paying you back for years, which feels more sustainable than burning cash in a straw fire on TikTok ads.

DR itself doesn’t hold any value. But it is a nice game chasing numbers. It feels productive, and I tell myself it helps with link exchanges in case one wants to try it out.

Publications

Along the way, two newspapers wrote about us

One of them a Sri Lankan newspaper we never reached out to. A nice reminder that things spread in ways you can't plan.

A few more articles are planned. Actually the interview for a third one is just completed and is scheduled for publication in a few days.

The Next Marketing Experiment

The next experiment is the one I'm most excited about.

We're sitting on more than 30,000 completed lessons, and we're analyzing that data to publish original research on attention spans and learning behavior.

If it lands, that's publicity money can't buy.

We sell in-app-subscriptions 7.99 USD for a monthly subscription, or 49.99 USD for a yearly.

Our costs are hosting, AI inference, and marketing. Backlinks as featured content.

Convex backend, react native app, node.js, and Vertex AI + Gemini API Key.

We use Google models, because their flexibility is immensely helpful: One API Key gives you LLM, top tier image, video, voice, embeddings, and all we could ask for.

We wrote the first line of code in December 2025 and launched on March 1, 2026.

Not even five months later we're at 10,000 downloads, 4.8 stars across 150 reviews, and just past 100 paying subscribers. We generated 3,000 USD in revenue in July 2026.

Crossing 100 felt bigger than any download milestone. A download is curiosity. A subscription means someone decided a thing we built at the kitchen table is worth their money every month.

Our nerdsip.com website is ranking at 3M views and 13k clicks with strong upward trajectory. We have a ton of articles on the first page of Google, but often score really low clicks in %. But in parts that's the new age of SEO with LLM search. We get mentioned in the AI citations a lot, so people don't end up on the page. And more than half of the queries are clear stuff that LLMs prompt and not a real human.

You can find NerdSip at nerdsip.com,in the App Store, and Google Play.

See you next time.

Thanks,
Jakob Jelling

I’m on X. Come out and play.