We built this because we worked inside the quality gap.

The formation gap that shouldn't exist.

Katherine Larsson spent several years as a process quality engineer at a battery cell manufacturer in Nevada — standing in formation rooms, reading SPC dashboards, watching shift managers hold the same post-mortem after every failed EOL audit. Yield dipped. Trays got flagged. Investigation started. By the time root cause was identified, another four shifts had run the same process, generating the same defect, quietly filling the scrap bins.

The data to catch those defects earlier was already there. Every formation cycler on the line was logging voltage, current, temperature, and capacity at sub-second resolution, for every cell, for the full 6–8 hours of cycling. The signal was legible to anyone who knew where to look in the dQ/dV spectrum. But nobody had built a tool to read it in real time and surface an alert before the cycle completed — the tooling that existed was built for post-process archival, not in-cycle inference.

In 2024, Katherine moved to Salt Lake City, connected with Dmitri Voronov — then building industrial time-series anomaly detection systems — and started Moldpathio to close that gap. We are not a general industrial IoT company that added a battery vertical. Moldpathio was built for one problem, in one stage of manufacturing, and the team's backgrounds are shaped entirely by that problem.

Our north star is simple: defects caught in the chamber, not at the dock. Every feature we build serves that constraint. We are not building a data lake, a manufacturing execution system, or an ERP integration layer. We are building the 40-minute early warning system that gigafactory formation teams need and have not had.