Most students are still figuring out their first to-do app tutorial when they enter college. I decided to build a company instead.
I'm Adhitya, a B.Tech student at IITRAM, Ahmedabad. In my first two years of college, I built and shipped 7 live, monetized AI applications under my indie studio Rewrite Labs — plus 7 open source libraries used by other developers.
No CS degree required. No internship at a big tech company. Just a laptop, free tiers of incredible tools, and an obsession with shipping.
Here's everything I wish someone had told me before I started.
1. The Stack That Actually Works for Solo AI App Builders
Before I talk about lessons, here's the exact stack I use across all my apps. I landed on this after painful trial and error and I haven't needed to change it since:
| Layer | Tool | Why |
|---|---|---|
| Frontend | React + Vite + Tailwind CSS | Fast dev, huge ecosystem |
| Auth | Clerk | 10-minute auth setup, handles JWTs |
| Database | Supabase | Postgres with a clean dashboard, free tier is generous |
| AI | Groq API | Blazing fast inference, free tier, llama-3.3-70b is incredible |
| Payments | Razorpay | Best option for India, subscription plans supported |
| Deployment | Vercel | One command deploy, serverless functions built in |
This stack lets me go from idea to live, paying app in under a week. Every tool has a free tier. Every tool has great documentation. There's no reason to use anything more complex when you're building solo.
2. Groq Will Change How You Think About AI Apps
Most people default to OpenAI. I did too, for about five minutes.
Then I tried Groq — and it felt like switching from dial-up to fiber.
Groq runs LLM inference on custom hardware called LPUs (Language Processing Units). The result? Responses that stream so fast they feel instant. For a chat app like my AI companion Nura, this is the difference between a product that feels alive and one that feels like it's thinking too hard.
The API is almost identical to OpenAI's, so migration is trivial. And the free tier is genuinely usable — not just a "taste."
If you're building any AI app today, start with Groq. You can always switch models later.
3. Ship Ugly. Polish Later. Seriously.
My first app took way too long because I kept polishing before shipping.
The brutal truth: nobody is looking at your app until you tell them to. That means every hour you spend pixel-perfecting before your first 10 users is an hour wasted.
My current process:
- Build the core feature loop in 2–3 days
- Deploy it live (even if it looks rough)
- Get 5 real people to use it
- Fix what breaks
- Then polish
The version of your app you ship first will look nothing like version 1.0. Stop treating the MVP like a portfolio piece.
4. Auth is Not Your Problem — Stop Building It
Early on I wasted two days building a custom authentication system. Session tokens, password hashing, email verification — the works.
Then I discovered Clerk and deleted all of it.
Clerk gives you:
- Email + social login
- JWT verification for your API routes
- Pre-built React components
- User management dashboard
Setup takes literally 10 minutes. It's free for the first 10,000 monthly active users. There is zero scenario where you should be hand-rolling auth as a solo builder in 2025.
Same principle applies everywhere — use Supabase instead of building your own database layer, use Vercel instead of managing your own server. Your job is to build the product, not the infrastructure.
5. Monetization Is Easier Than You Think (If You Start Early)
I put a paywall in my second app. It felt bold, almost arrogant — who would pay for something a college student built?
Turns out: people who find it valuable.
Here's what I've learned about pricing AI apps as a student:
- Start charging from day one. A free plan is fine, but have a paid tier. It forces you to build something worth paying for.
- Subscriptions beat one-time payments. Recurring revenue means you can predict growth. Razorpay makes this straightforward with their subscription plans API.
- The paywall is actually a feature. It filters out users who aren't serious. Your paying users give better feedback and stick around longer.
Don't wait until your app is "ready" to think about money. The business model should be part of the design from the start.
6. Open Source Is Your Best Marketing
While my apps are private (protecting the IP), I extracted reusable pieces into 7 open source libraries — things like:
-
react-premium-gate— Razorpay subscription paywall components -
react-macro-rings— animated SVG nutrition progress rings -
react-toast-native— lightweight toast notifications -
groq-chain— Python LLM chaining without LangChain -
llm-router— routes prompts to the right LLM by complexity and cost
Each library is a tutorial disguised as code. Developers find them, use them, and discover who built them. It's the most authentic form of marketing there is — you're not selling anything, you're just being useful.
If you're building apps, look for the pieces you keep rebuilding and open source them. It builds your GitHub profile, your reputation, and your network simultaneously.
7. The Thing Nobody Tells You About Building in College
Everyone focuses on the technical side. The harder part nobody talks about is the mental game.
You'll have weeks where nothing works. Where a bug takes three days to find and it's a single missing await. Where you ship something and nobody cares. Where your college workload collides with a critical bug in production.
What keeps you going isn't motivation — motivation is unreliable. What keeps you going is having built the habit of showing up and shipping anyway.
Some things that helped me:
- Build in public (even quietly). Writing about what you're building creates accountability.
- Ship something every week, even if it's small. Momentum compounds.
- Separate your identity from the product. Your app failing doesn't mean you failed. It means you learned something that cost you nothing but time.
The builders who win aren't the smartest. They're the ones who kept going when it stopped being fun.
What I'm Building Next
Rewrite Labs is still growing. The app suite is expanding. The OSS library count is going up.
If you're a student reading this wondering whether to start — start. The tools have never been cheaper, the knowledge has never been more accessible, and the market has never been more open to AI products.
You don't need permission. You just need to ship.
Follow along — I'll be posting technical deep-dives, build logs, and honest lessons from the trenches right here on Dev.to.
Top comments (0)