Data Flows — The Origin

Why this case study exists

Not to showcase screens. To show where TX-1 came from.

Data Flows is a cloud-based ETL tool for supply chain data transformation — a visual node graph interface that lets analysts string together data cleaning and transformation operations without writing code. Built on top of a legacy desktop application called Data Guru, it handles the kind of messy, time-intensive data preparation work that sits at the foundation of every supply chain model.

I spent years designing and refining it. I had ideas that didn’t get built. I watched analysts use it in ways we hadn’t anticipated, work around limitations we hadn’t solved, and spend hours on operations that should have taken minutes.

The insight that changed everything

The visual node graph was the right idea for the wrong reason. We built it because it was more accessible than writing SQL. But the real problem wasn’t that analysts couldn’t write SQL. It was that they were manually making decisions — about data quality, about transformation logic, about field mappings — that a sufficiently intelligent system should be making for them.

The interface was fine. The paradigm was wrong.

That insight is the direct origin of TX-1’s manifest-driven schema inference and agentic auto-fix loop. The question I kept asking while building Data Flows — “why is a human doing this?” — became the design principle that TX-1 is built around.

What the work demonstrated

A decade of enterprise UX work produced specific capabilities that don’t show up in screen designs: understanding how analysts actually think about data quality problems, knowing where the cognitive load in complex workflows is genuinely unavoidable versus artificially imposed by bad design, and developing the domain knowledge to talk credibly to engineers, data scientists, and supply chain specialists simultaneously.

That’s the foundation TX-1 is built on. Not the screens. The understanding.