We have been designing connected hardware for close to ten years, starting with Espressif's original ESP chip and the small devices we built around it. Since then the work has covered logistics devices, trackers, and controllers, from board layout and enclosure through firmware, production, and the services behind a deployed fleet.
Hardware has been part of this practice for close to a decade. It began with Espressif's original ESP chip, which at the time was the cheapest way to put a radio and a microcontroller into something small, and it continued through a long run of devices built for real operational use: logistics hardware that had to survive a warehouse, trackers that had to report from places with poor coverage, and controllers that had to actuate something physical without being babysat. None of that was a side project. It was shipped, deployed, and supported, and most of it is still running.
We do PCB layout and enclosure design in house rather than handing a schematic to a stranger and hoping. The two are not separable problems: antenna placement constrains the board, the board constrains the enclosure, and the enclosure constrains how much battery will fit and how warm it is allowed to get. A team that only writes firmware will discover those constraints after the boards are fabricated, which is the expensive moment to discover them. Designing the board and the housing together is how the device ends up the size the product actually needed.
Battery life is decided in firmware long before anyone measures it in the field. A device that wakes on a fixed interval, holds its radio open waiting for an acknowledgement, and writes verbose logs to flash will drain a cell in days. The same hardware running firmware that batches its readings, sleeps between them, and transmits only when it has something worth sending will run for months. We write firmware for microcontroller-class, battery-powered devices that sample sensors, report telemetry, and accept updates after they have shipped. The power budget is a constraint we design from rather than a number we discover at the end.
A device that works on a bench is a long way from a device that can be built a few thousand at a time. Parts go out of stock, tolerances that did not matter in a run of ten start mattering in a run of ten thousand, and test and provisioning have to happen on a line rather than at a desk. We take designs through large scale production, which means the decisions made early are made with volume in mind instead of being revisited expensively later.
A device is only as useful as the system receiving what it sends. We build and operate the ingestion endpoints, the storage, and the APIs that device data flows into, on infrastructure we configure and maintain ourselves, following the same practice described in infrastructure independence. There is no third-party IoT platform sitting between your fleet and your data. That means no pricing change you did not plan for, no deprecated API that silently bricks a feature, and no breach notification arriving on a vendor's timeline instead of yours.
Nearly all of the hardware we have built over the last decade was built for someone else. MaverickTracker, a GPS tracker for dogs, is the first device we will sell under our own name, and it is launching soon. It pairs a collar-mounted device with an application showing live location and activity, and it is designed to run for months between charges rather than days. Selling our own device means the board, the enclosure, the firmware, the ingestion service, the application, and every support call land on us, which is a useful discipline to submit to before asking a client to trust us with theirs.
Connected products fail at the seams. The firmware team blames the backend, the backend team blames the firmware, and the person who bought the device waits for someone to take responsibility. We build both halves, so there is no one to point at. If you are building something that measures the physical world and reports on it, contact us and we will tell you honestly whether it is work we should take.