The Nine-Month Impossible: How I Rebuilt Airstream's Connected Platform Without Stopping the Train

Nine months. That’s what I had to scale Airstream’s connected vehicle platform from one trailer model to several — including a high-profile Pottery Barn collaboration that had already been announced publicly.

The catch: the platform we were building on was fundamentally broken. Not “needs some refactoring” broken. “Would take 2-3 years to fix properly” broken.

I had to figure out how to ship anyway.

Inheriting the debt

When I came in, the platform had three problems that compounded each other badly. We were running on yacht-grade C-Zone hardware — expensive enough that expanding to lower-priced trailer models was mathematically impossible. Our backend was built on Parse, which had been deprecated, creating constant state management chaos between the mobile app and our MongoDB database. And we had no hardware abstraction layer, which meant every new RV model would require starting over from scratch.

The honest assessment: properly rebuilding this would take a decade to scale across our full product line and $1.5M a year to maintain in the meantime. That wasn’t a technical problem anymore. It was a strategic one.

The hardest conversation

At a quarterly business review, I made the case for abandoning our current platform entirely and pivoting to a new architecture. The room was not enthusiastic. We’d spent 2.5 years building what we had. Dealers were already navigating warranty issues on it. And now I was standing there saying we should walk away from it.

The sunk cost argument is always the hardest one to fight — not because it’s logically sound, but because it feels emotionally true. Nobody wants to believe the last two years were a wrong turn.

I’d done the math, and I walked them through it: staying the course cost more, took longer, and still wouldn’t get us where we needed to go. The pivot was the only path that could hit our nine-month window.

The breakthrough came from an unexpected place — our European sister company had faced nearly identical constraints and had already built a modular architecture with a low-cost control unit that solved exactly our problem. Instead of reinventing it, we adapted it. Hardware costs dropped 85%. The path to scaling across our full product line opened up.

When the chip shortage hit

Of course, nothing stayed clean for long.

Midway through the project, the global semiconductor shortage delayed our connectivity modules by eight weeks. We had ten weeks left before our prototype deadline.

We negotiated with our consulting partner to swap frontend engineers for embedded specialists, built a virtual testing environment to simulate the chipset behavior while we waited, and secured two pre-production modules from our vendor for priority testing. We pushed the prototype milestone two weeks and protected every downstream deadline.

That virtual testing environment wasn’t just a workaround — it became a permanent asset that cut future integration time by three weeks minimum. Some of the best infrastructure I’ve seen come out of crisis mode.

The AT&T ultimatum

Just as the chip situation stabilized, AT&T told us we had eight weeks to agree to an exclusive retail partnership or they’d cut our service. This was the same provider whose coverage gaps were already frustrating customers trying to connect in remote areas. Their ask made no sense for our users.

So we left.

I led a parallel workstream: evaluated providers, landed on T-Mobile for coverage — right before their Sprint merger closed — updated router configurations and APNs with our hardware vendor, built in-app SIM replacement guidance with UX and engineering, and coordinated distribution of new SIMs to existing customers with clear instructions.

Zero downtime. Over 90% SIM activation in the first month. 20% reduction in data costs. And customers in remote areas could finally get a signal.

What shipped

The Pottery Barn edition launched on time for MY22. 700+ pre-orders. The modular architecture we built became the foundation for the rest of the product line. The persistent warranty issues that had been requiring dealer visits disappeared.

The thing I remember most isn’t the launch — it’s the quarterly review where I had to argue against two and a half years of work and convince people the right move was to start over. That’s the decision everything else depended on. Getting that one right mattered more than any of the execution that followed.