Dutch AI Lets You Search Company Data Like Google

Imagine asking your company's data a question in plain English and actually getting an answer. That's the breakthrough Madelon Hulsebos just made possible.

The Dutch researcher noticed something frustrating during her years as a data scientist. Highly paid specialists spent most of their time doing the same boring tasks: cleaning spreadsheets, linking datasets, and hunting for information that already existed somewhere in the company's files.

"We don't know what we don't know," says Hulsebos, who leads the Table Representation Learning Lab at the Dutch Centrum Wiskunde & Informatica. Her team of 11 researchers is solving a problem that sounds simple but stumps most organizations.

Companies have mountains of organized data in databases and spreadsheets. The information exists, but finding it requires knowing exactly which system to search, what each column is called, and how to write complex database queries. One department might label something "customer ID" while another calls it "client number," and suddenly your search comes up empty.

Hulsebos developed something called "table representation learning." Instead of just searching databases, her AI actually understands what the data means. It's like the difference between a student memorizing facts and one who truly grasps the concepts.

The five-year DataLibra project runs through 2029, funded by a National Growth Fund grant to keep top AI researchers in the Netherlands. The goal isn't just research but building actual tools companies can use tomorrow.

"It should be as simple to query data within your organization as it is to perform a Google search," Hulsebos explains. No programming knowledge needed, no database expertise required.

The Ripple Effect

This breakthrough could reshape how millions of people work. Data scientists famously spend 80% of their time preparing data and only 20% actually solving problems. Hulsebos wants to flip that ratio.

"We want to automate that 80% as much as possible, so data scientists can focus on critical aspects of problems, such as ethical questions," she says. That means faster insights, better decisions, and data experts finally doing the creative thinking they were hired for.

But the impact goes beyond data teams. Marketing managers could find customer trends without waiting for IT. Hospital administrators could spot patterns that save lives without hiring consultants. Small businesses could compete with big companies that can afford data science teams.

Hulsebos remains realistic about the hype. She's tested the fancy AI tools companies are selling, and the success rate is often zero. "It all sounds wonderful, but to actually get there, we still have much work to do."

Her team already completed a project for the United Nations, proving the concept works in real-world conditions. The key difference from existing tools? Her system explains why its answers are right, providing the transparency people need to trust the results.

The researcher emphasizes that full automation isn't the goal because nobody should blindly trust what AI tells them. Instead, she's building tools that augment human judgment, making expertise accessible to everyone while keeping humans in control.

By 2029, asking your company's data a question might feel as natural as Googling something today.