Building My MicroSaaS
And Everything’s Falling Apart (In the Best Possible Way)
🧠 It Seems Like Every Dev Is Building a SaaS These Days
Just open Twitter or LinkedIn, and there’s always someone launching a landing page, a “Sign up” button, and a shiny new freemium plan.
And I found myself wondering: is this the new formula for success?
But instead of watching from afar, I decided to find out for myself — and started building my own microSaaS.
The result?
Everything’s going wrong… and that’s been incredibly right.
💡 The Idea: Solving My Own Problem
Every time I hosted a party at home, chaos would follow — especially when it came to splitting costs among friends.
Spreadsheets, WhatsApp groups, and manual math.
No existing product did it well.
We even had an Excel sheet that kind of worked: it helped us figure out who should pay what.
In a barbecue, for example, vegans or non-drinkers shouldn’t be charged for meat or alcohol, right?
So we manually adjusted everything using Excel formulas — plus Google Forms to track who drank, who didn’t, and who had food allergies.
That pain point became the seed for my MVP:
a simple, straightforward, and smart app to manage and split costs automatically.
⚙️ The Technical Stack
I wanted to build it the right way, as a developer, not a “maker in a hurry” armed with too many AI tools.
But I also learned how to make the modern ecosystem work for me.
Here’s what powers the project today:
TypeScript + Node.js: the foundation of the app’s core, built from scratch with SOLID principles and automated tests.
Supabase: a backend-as-a-service with managed Postgres, authentication, and edge functions, great for fast MVP validation without skipping security.
PostgreSQL: my main database, versioned with migrations and seed scripts.
Edge Functions (Supabase): lightweight APIs running close to users with refined security logic; I only use them for integrative flows — the real backend logic I prefer writing myself in TypeScript.
Redis (local cache): lightweight caching to avoid unnecessary reads.
AbacatePay: a custom payment-splitting module I’m building — when an event ends, it generates a Pix code so guests can pay the organizer directly.
Resend: email delivery with my own custom domain, minimal and clean. Guests receive event updates automatically — it was one of the first features I integrated.
GitHub Actions: full CI/CD pipeline with Playwright E2E tests and unit coverage using Jest.
AWS + Serverless + S3: used for parallel integrations and asset deployment.
Lovable was just the kickstart — I used it to generate the initial template (auth, base layout, and config), and then jumped into real code.
That setup alone saved me days and let me focus on what truly matters: security and architecture.
I also used Lovable as my UX starting point — and honestly, for backend devs like me, it’s a lifesaver.
If I had to design the whole flow from scratch in Figma, it would’ve taken me ages (and still looked bad 😂).
🔐 Where Things Are “Going Wrong”, and Why That’s a Good Thing
Security: I noticed some AI-generated code exposing sensitive payload data. I had to manually revise authentication, tokens, and roles. It’s not the end of the world — any AI output can hide vulnerabilities. Instead of wasting tokens asking AI to fix it, I built my own filters, validations, and tests.
Costs: Pricing is harder than scaling. I’m still experimenting with subscriptions, event-based pricing, and free tiers.
Deployment: Configuring GitHub Actions with Edge Functions and end-to-end testing took longer than the backend itself.
Validation: I launched it only to a closed group of friends for early feedback. They’ve basically become my QAs — which is great, but also revealed plenty of UX mistakes (part of the process!).
And honestly, I’m learning more in this chaos than in any sprint I’ve done.
🧭 A Quick Career Detour
Through this process, I realized how similar building a product is to working for international companies.
You need to think like an owner, communicate clearly, and deliver value autonomously.
That’s how I came across Coders, a mentoring initiative that helps devs structure their international careers strategically.
I joined one of their 1:1 sessions — and it was the kind of talk that shifts your entire perspective.
If you’re curious, I really recommend checking them out and booking a free 60-minute session:
👉 Coders Mentorship Program
❓Is Everything Really Going Wrong?
Building this microSaaS isn’t about launching a startup.
It’s about going back to the roots of software engineering — understanding the product end to end, from database to deployment, from UX to security.
Every mistake has taught me more about architecture, automation, observability, and product thinking than most corporate projects ever did.
And I finally get the hype: it’s never been easier to bring wild ideas to life — using code intelligently to create real value.
Still, I do worry about the quality of many new SaaS products out there.
Some are built by people who skip testing or ship insecure code that leaks user data.
That makes me wonder if this new “build fast” era is really for everyone.
If everything seems to be “going wrong,” it’s actually a good sign —
because it means I’m truly learning.
I had a similar realisation. Keep it up! Have you worked out how to market?