The Blue Ocean Hiding in Plain Sight

For most of my career, I've worked with tabular data.

As a data scientist, that's where the real value has always been: structured, business-critical data that powers decisions across every enterprise. Revenue forecasting, risk modeling, churn prediction, pricing. Behind all of it is tabular data. And yet, looking back on the past ten years, progress in this space has been surprisingly stagnant. Models improved incrementally and tools became more convenient, but nothing fundamentally changed.

Then came the explosion of LLMs.

Like many, I was excited. LLMs made me faster, more productive, and more creative in my day-to-day work. They changed how we interact with information. But the more I used them, and the more I observed how companies were adopting them, the more I felt a growing disconnect.

Something was missing.

Despite the huge promise of LLMs, they didn't seem to justify the scale of investment flowing into them. I witnessed this first-hand working in a startup in the Bay Area and collaborating with a wide range of businesses. Budgets were flowing into GenAI, especially amongst companies that had missed the first wave and were rushing to catch-up – often without fully understanding where the real value would come from. There was urgency, but not always clarity.

Billions of dollars were being poured into GenAI, but when you looked closely at most companies, the core drivers of value hadn't changed. The most critical decisions were still powered by tabular data, and that space remained largely untouched by this new technology.

And yet, beneath that surface, something more important was starting to take shape.

When I came across Fundamental and their work on Large Tabular Models (LTMs), everything clicked.

For the first time in years, it felt like we were at the beginning of a true paradigm shift in predictive algorithms, this time for tabular data. Not just incremental improvements over XGBoost or deep learning variants, but the first real breakthrough where a new class of models starts to outperform traditional machine learning approaches.

It reminded me of where LLMs were around 2020.

Back then, the early signals were there. Models were just starting to outperform previous approaches on certain tasks, but most people underestimated what was coming. The progress didn't look revolutionary yet. But what mattered wasn't where the curve was. It was the shape of the curve. It was exponential.

We all know what happened next.

What we're seeing today with LTMs feels remarkably similar. We're at the very beginning of that curve, where the gains are just becoming visible, but the long-term potential is massive. A new frontier. A blue ocean.

What makes this moment even more compelling is that it's happening where most of the real enterprise value actually sits.

LTMs have the potential to redefine how we approach prediction, decision-making, and value creation across industries. Not through incremental gains, but through a fundamental shift in capability: unlocking the full potential of the data companies already have.

That's why I joined Fundamental.

It felt like one of those once-in-a-lifetime opportunities, where the timing, the technology, and the people are perfectly aligned. A true inflection point, with the right team to make it happen.

Those are the moments you don't pass on. I'm humbled to be a part of it.