Alessandro Sillato

Power BI is not enough (anymore)

For years, the answer to any business intelligence question was the same. You need dashboards? Power BI. You need reports? Power BI. It became the default choice. Safe, familiar, everyone uses it. The question stopped being which tool should we choose and became how do we use Power BI.

But something changed. I think we missed it.

Today, if you need a custom button in Power BI, you cannot do it. If you want to connect your dashboard to something Power BI does not support, you have to find a way around it. If you need to export data in a specific format, the tool does not help you. These are not small things. They add up.

The real problem is not Power BI itself. Power BI works fine. The problem is that Power BI was made in a different time. It was made when building your own solution was expensive and difficult. When you needed many developers and many months just to build something basic. That time is gone now.

AI coding assistants changed things. Claude, GitHub Copilot, and others. Today I can describe a dashboard and get React code that works. I can describe an API and get a backend that works. The cost of building your own solution went down a lot. What took weeks now takes days. What took days now takes hours.

So I asked myself: what if I stop accepting the limits of a commercial tool. What if I build something myself.

This is what I am testing now. It is not finished. It is not perfect. It is a proof of concept. But it works and it is interesting.

The stack is simple to understand. Data models are made with Fabric. Fabric handles the hard part of transforming and modeling data. Then a backend service connects to Fabric and works with the gold layer of the medallion. The backend also does something that commercial tools do not do well: it manages a cache layer where you can control when the data refreshes. It is like a semantic model, but you control it completely.

Above that is React. Not a drag and drop tool. Real code. Real components. This is where you see the difference. I can build what the business needs. A custom chart? Yes. A specific export format? Yes. An integration with another system? Yes. There are no limits.

This approach has costs. You own the code. You own the servers. You own the maintenance. Power BI takes care of these things. But you also accept the limits. And now, those limits matter more.

The good part is this is not crazy. I am not building everything from zero. Fabric handles the data side. Claude helps me build fast. React is stable and known. I am putting together pieces that already exist and work. Five years ago, this would be too risky and too expensive. Today it is a good bet.

What bothers me about most BI tools now is they think you will stay inside them forever. Power BI, Tableau, others. They were made thinking that you will accept the limits. But now things are different. Building yourself is cheaper and safer than staying inside the limits.

The question is not if commercial BI tools will disappear. They will not. Most teams do not need custom dashboards. Most teams like knowing they have a tool that works and has help. But for teams with real problems to solve, teams that have the skills, teams that want to try something different, things have changed.

I do not know yet if this test will become real or if it will stay a proof of concept. That depends on if it actually works better. But I know one thing. Now you can choose to build. For the first time in a long time, choosing to build is not the wrong choice.

That change means something about where we go from here.