What happens to your smart home when the company disappears

I’ve written twice now about gear that stopped doing what the buyer thought they paid for. The Ring doorbell that turns into a chime on day 31. The Harmony remote that Logitech quietly stopped making. Different companies, different failures, and people keep asking me the same follow-up question: how do I avoid this next time?

The honest answer isn’t a brand. It’s a question you can ask in the store.

Harmony is the cleanest experiment we have

Logitech walked away from the universal remote category. No new hardware, no meaningful updates. That’s about as close to “the company disappeared” as a large company gets.

And my house is still full of Harmony remotes that work every day. Press the button, the activity runs, the TV and the receiver and the streaming box all do the right thing. Nothing about Logitech’s decision reached into my living room and broke that.

But if I want to add a device to one of the older units, or change what a button does, that’s a different story — the older Harmony software lost server access in the spring of 2025, and reprogramming those units just stopped working.

Look at what survived and what didn’t, because the line between them is the whole point. The activities run locally, on the hub, so they kept running. The part that needed to phone home to Logitech to make a change is the part that died with the servers.

That’s the split. Everything else in this post is that same split, applied to gear you might be about to buy.

The question worth asking

Before you buy any connected device, ask this:

If this manufacturer turned off their servers tonight, what would this thing still do tomorrow morning?

Not “will they turn them off” — you can’t know that, and neither can the salesperson. The useful version assumes it happens and asks what’s left.

For a Harmony hub, the answer is: most of it. For a camera whose video only exists in that company’s cloud, the answer is: nothing you’d want. For a doorbell whose recording lives behind a subscription, you already know the answer, because it’s the same answer you get on day 31 without ever needing the company to fail.

Most products land somewhere in between, and almost nobody selling you one has ever been asked to say where.

Why I care about this more than the average installer

A lot of our business is ripping out expensive, dealer-locked automation platforms. Savant, Crestron, Control4 — real systems that work, sometimes beautifully. The reason people call us to remove them usually isn’t that they broke. It’s that changing anything means calling a dealer.

Same shape as the server problem, one layer up. The device works. The ability to change it lives somewhere you don’t control. And whether that somewhere is a cloud service or a company with a truck, you find out how much it mattered at the exact moment you want something different.

The part where I don’t oversell the alternative

The clean answer to all of this is to run the smart home yourself, locally, on something like Home Assistant. Nothing to shut off, because there’s no one running it but you.

I’ll be straight about the cost of that, because I’ve said it here before: for most of our customers, Home Assistant is a bridge too far. It is genuinely more work. You are trading a company that might leave for a system you have to maintain, and that is a real trade, not a free upgrade. Plenty of households should not make it.

There’s also a middle path most people miss, which is to buy things that keep working without the company and simply not care about the rest. Local recording. Standards like Z-Wave, Zigbee, Matter that more than one manufacturer speaks. A hub that runs its logic on the hub. None of that requires you to become a hobbyist. It just requires asking the question above before you pay, instead of finding out afterward.

What I’m not going to tell you

I’m not going to predict which company is next. I don’t know, and anybody who tells you they do is selling the alternative.

What I can tell you is that the two failures I’ve written up here weren’t bad luck. One was a company leaving a category, the other is a business model working exactly as designed, and in both cases the customer’s exposure was decided long before — at the moment they bought a thing whose important parts lived on somebody else’s computer.

That part you can check. It’s the only part of this you control.