A manifesto for runtime intelligence

For thirty years, process mining has been the most important operational discipline most people have never heard of, and it runs the digital backbone of every Fortune 500.

Process mining is the discipline of recovering how work actually flows through a system by analyzing the data it produces. It takes into account every path, including all the exceptions and edge cases.

Meanwhile, physical operations are drowning in data and starving for insight. We have more operational telemetry than ever before, but less ability than ever to connect it to the decisions that move your business. The data is there, the model that connects it to dollars and sense, does not.

The great part about this, is that physical operations are already digital. Every robot runs software, every machine streams data, every autonomous system is just a computing system that happens to move things in the real world.

The only way to truly understand a complex digital operation is to recover a model of how work actually flows through it. Not from documentation or dashboards, but from the data the system itself produces. That discipline transformed how we manage digital systems, but it has never been applied to the physical world. We believe it's the most consequential missing layer in physical AI.

Closing it means rebuilding process modeling for the physical world; where data is continuous, processes evolve dynamically, identity is probabilistic, and the clock runs in milliseconds, not minutes.

Continuous understanding over batched measurement. A system that changes every second cannot be understood once a day.

Recovered models over authored specifications. Autonomous systems don't follow scripts. The only honest model is one mined from what the system actually does.

System intelligence over dashboard management. The goal is better operations, not better dashboards.

Closed-loop feedback over open-loop reporting. Measurement must feed back into the system at machine speed.

While there is value in the items on the right, we value the items on the left more.

Process mining gave the digital world a way to see itself clearly. Physical AI deserves the same. Built for systems that move, learn, and change while you're watching.