🛤️ When building connected products, two things have to go right

The first is architecture, decided partly up front: network topology, the protocol between processors, how function is partitioned across boards, real-time constraints, the data model, how firmware gets updated in the field, and how operations are logged and traced. These decisions are cheap this month and expensive in six months, because everything built afterward assumes them.

The second is the learning loop. No amount of planning removes the need to build, measure, and find out; what we control is what each round costs. The goal is to tool up so prototype and test cycles run quickly in both hardware and software: boards designed for fast bring-up and instrumentation, firmware that builds, flashes, and exercises itself automatically, and a test fixture or simulation that continually verifies the system. A team that invests in that loop early can do more loops, more efficiently, and the learning compounds. Time is the scarce resource, so the loop is where we buy it back.

The first is the fruit of experience, the second the ingredient for innovation.