FinanceHub #1: Cuando un MVP deja de ser suficiente

FinanZen #1: When an MVP Ceases to Be Adequate

Introduction

Some time ago, alongside a colleague, we embarked on developing FinanceHub. The idea was straightforward: create an application to help us better understand our personal finances. Like many projects, we wanted to validate the concept as soon as possible, so we began using a no-code platform. And it worked. In a short time, we had a usable application.

We could record incomes, expenses, and visualize part of our financial information. For an MVP, it was exactly what we needed. However, a question arose. As time went on, we started imagining new functionalities. More business rules. Greater personalization. An API of our own. Integrations. Better security. More control over the infrastructure.

And there came a question that many developers have probably asked at some point: Should we continue building on the current platform or is it time to create our own technological base?

The Decision

Ultimately, we decided to take the leap. Maintain the idea. Maintain the product. But rebuild the platform entirely. My challenge was to lead this technical migration: switch from a no-code solution to a backend developed with Spring Boot, PostgreSQL, and an architecture designed to grow. The curious thing was that I thought the hardest part would be writing the new system.

It wasn't. The hardest part was deciding how to build it. Because when you stop depending on a platform, all the decisions become yours: authentication, the architecture, the infrastructure, security, contracts, the database, migrations, deployments, observability. Everything. The most important decision before writing a single line of code was to stop coding and start designing.

Thus, a methodology based on specifications was born. Before asking Claude Code to do any task, I defined: the architecture, the data model, the OpenAPI contract, the project rules, and a plan divided into small verifiable phases. That process turned out to be much more important than any prompt. What surprised me the most During this migration, I realized something.

AI can write a lot of code. But it cannot decide what system you want to build. That is still the work of the engineer. And the more complex the project, the more important it becomes to clearly define rules before starting. What's next In the upcoming articles, I want to share how we made those decisions. Not to teach a particular framework, but to show how we are approaching the construction of a real product from an engineering perspective.

Because at the end of the day, FinanceHub stopped being just an application for managing finances. It became a laboratory where we are learning about architecture, cloud, AI agents, and software design.