From a legacy desktop SaaS to an AI-first workflow — and a financial ecosystem built for the next generation of Brazilian entrepreneurs.

Context

When I joined yampa, the company was fully dependent on 4blue — its parent company — for revenue and direction. MRR was R$80k/month. The product had been built by a single developer, which meant no clear information architecture, no scalable structure, and new features being added without connection to related areas of the platform.

The culture was traditional — waterfall processes, no product thinking, no design system. Transforming it into a startup mindset was as much a challenge as redesigning the product itself.

The business model was also unclear. There was only one type of subscription, varying only by period: monthly, quarterly, semi-annual, and annual. No upsell strategy. No differentiation between customer segments.

Phase One — Business Strategy

Before touching the product, I focused on the business model.

Drawing from experience with other SaaS products, I proposed creating multiple subscription tiers — each with specific features — to enable a clear upsell strategy. The goal was to grow revenue without necessarily growing the number of subscribers.

The team was convinced through a straightforward argument: sell more and sell better. Within six months, MRR had grown from R$80k to R$200k. By the time I left, it had reached R$370k — without a proportional increase in the subscriber base.

Phase Two — yampa 2.0

With the business model stabilized, we faced the next problem: retention.

Customers coming from 4blue's courses stayed. Customers from open market acquisition didn't. The product wasn't self-explanatory, and it had no attractive experience for users arriving on mobile.

We decided to build yampa 2.0 — a complete redesign with a mobile-first approach, cleaner information architecture, and a product that could guide users without requiring a course first.

I led the full product and design process for two years. Discovery, research, UX, UI, design system, and coordination of the development team.

Then, just before launch, a more senior developer joined the team.

Within weeks, he identified critical structural problems in the codebase — decisions made by the previous developers that would make the system impossible to scale. The backend and database were fundamentally flawed. The frontend couldn't be salvaged either.

Two years of work couldn't ship.

It was frustrating. The product and design work was solid. But without a viable technical foundation, there was no argument to make.

So I had to reinvent.

Phase Three — The AI-First Shift

The constraint became the strategy.

We had a working backend and database. We needed a completely new frontend — fast. The decision was collective, born out of necessity: rebuild everything with AI-first vibe coding.

This wasn't our first contact with AI. During yampa 2.0, we had already started using AI for specifications, research organization, design system documentation, and PRD creation. But the product's poor architecture meant AI got lost in the complexity. We were using it as an assistant, not as a foundation.

The pivot changed everything.

Starting from scratch with a well-structured codebase, AI stopped being a helper and became the workflow itself. Designers and PMs were no longer handing off to developers — they were delivering production-ready frontend code directly. Some features that didn't depend on backend integration shipped complete, without developer involvement.

The entire frontend was rebuilt and improved in under two months.

Not perfect — quality is still being refined. But fast, functional, and built on a foundation that can actually scale.

Impact

People

Many people came and went through this project. A few stayed long enough to grow with it — and that growth was the best part of the work.

Lauren, Silas, and Matheus showed up every day and became sharper, more resilient collaborators through every pivot and restart. Watching that evolution was worth more than any launch.

Bruno and Rapha held the strategic direction through the hardest moments. The decisions weren't always easy, but they were always made.

To everyone still building yampa — good luck with what comes next.

Learnings

The best product work I've done at yampa wasn't the 2.0 — it was what happened after it was cancelled. Being forced to let go of two years of work and find a better path forward taught me more about product leadership than any successful launch ever could.

AI doesn't save a poorly structured project. During 2.0, we tried to use AI inside a broken system and it got lost. The real unlock came when we built the right foundation first — then let AI do what it actually does well.

Speed and quality are not opposites. Rebuilding the entire frontend in under two months with AI wasn't a shortcut. It was a different way of working — one where design decisions are validated faster, iterations happen in real time, and the gap between concept and shipped product nearly disappears.