Interoperability Hell: Why Your Smart Devices Still Won’t Talk to Each Other
You replaced your Wi-Fi router on a Saturday morning. By Saturday afternoon, the hallway lights had stopped answering, the robot vacuum was “offline”, and the smart plugs — all nine of them — were waiting patiently in an app you deleted two phones ago, under an account you don’t remember creating.
Nothing was broken. Every device still worked exactly as manufactured. They just no longer agreed on who was in charge, which cloud they answered to, or which network they were supposed to be on.
That’s interoperability hell: not a missing feature but a maze of walled gardens, half-adopted standards and abandoned clouds, built one reasonable-sounding business decision at a time. And in 2026 — three years after Matter promised to fix all of this — your devices still talk to their manufacturers far more fluently than they talk to each other. 🧩
How We Built the Maze
The maze wasn’t designed. It accreted.
The first smart homes ran on purpose-built mesh radios — Zigbee from 2004, Z-Wave alongside it — and if you’ve read my Zigbee vs Z-Wave piece you know those two never spoke to each other either. But at least they were standards: a Zigbee bulb from one brand could, in principle, join another brand’s hub.
Then came the venture-cloud decade. From about 2014, the cheapest path to a smart product was a Wi-Fi chip, a phone app and a rented cloud backend. No hub to build, no alliance to join — and no interoperability whatsoever, because the product wasn’t really the plug on your wall. It was the pipeline from the plug to the vendor’s servers to the vendor’s app. Multiply that by a few thousand startups and every gadget in your house became its own tiny, sealed ecosystem with its own account, its own firmware schedule and its own privacy policy.
Here’s the part most people get wrong: the radios were never the hard problem. Radios can be bridged — a $40 hub has spoken Zigbee, Z-Wave and Wi-Fi at once for a decade. The real fragmentation lived one layer up, in data models and accounts. One vendor’s “light” has brightness; another’s has scenes, effects and schedules; a third’s is a switch pretending to be a light. There was no shared answer to the simplest questions — what is a device, what can it do, who’s allowed to ask — and no commercial reason to agree on one, because agreement makes customers portable.
So the honest name for the era isn’t fragmentation, which sounds like an accident. It’s partition. Everyone built a wall because everyone else was building one, and the customer’s house was the land being divided.
The Walled Gardens
Onto that chaos, three companies imposed a kind of order — each with a wall around it.
Amazon’s Alexa, Google Home, and Apple’s HomeKit became the meta-layers: certify your device through their “Works with” programme, and it appears in their app, answers to their assistant, joins their automations. For users it genuinely helped — one app instead of nine. For vendors it meant certification fees, review queues and three separate integrations to maintain.
The walls are the business model. Each garden wants your next purchase to be easier inside than outside. Buy an Apple-friendly camera and HomeKit Secure Video makes leaving expensive; build your routines in Alexa and the routines are the lock-in. Features get gated too — the same bulb often does more in its maker’s app than through any assistant, because the assistant integration exposes the common subset and the maker keeps the party tricks for itself.
And multi-garden homes lose twice. Run Alexa in the kitchen and HomeKit on your phone, and each sees a different half of the house, at different levels of detail. The tollbooths don’t interconnect; you’re the bridge.
Matter: The Fix That Half-Arrived
Matter — the 2022 alliance of Apple, Google, Amazon, Samsung and the Zigbee old guard — attacked the right layer. As I wrote in the Matter & Thread piece: Matter is the language, a shared data model over IP, with local control and multi-admin so one device can serve several ecosystems at once, no cloud round-trips required. Genuinely the correct design.
So why doesn’t 2026 feel fixed? Three gaps.
Device coverage arrived in slow motion. The spec launched with the basics — bulbs, plugs, locks, thermostats, sensors. Robot vacuums waited for Matter 1.2 in late 2023; energy management, EV chargers and water heaters for 1.3 in 2024; batteries, solar and heat pumps for 1.4 at the end of that year. Cameras — the most-requested category in the entire smart home — only entered the specification at the end of 2025, and certified camera products are still trickling out as I write.
The standard is a floor, and vendors live on the ceiling. Matter deliberately specifies the common denominator. A Hue bulb’s gradient effects, a vacuum’s room-by-room maps, a lock’s guest schedules — much of that stays in the vendor app, because that’s where differentiation (and data) lives. “Works with Matter” is true and still not the whole truth.
And certification costs real money — membership plus per-product fees that a big brand shrugs at and a small one doesn’t. The standard meant to level the field has a paywall at the gate.
The Border Router Tangle
Thread, Matter’s favourite radio, brought its own flavour of fragmentation. A Thread network needs a border router, and for years every ecosystem shipped its own — a HomePod here, a Nest Hub there, an Echo somewhere else — each spinning up its own Thread mesh with its own credentials. Three border routers, three parallel networks, devices scattered across them. The mesh protocol designed for unity, partitioned by brand.
Thread 1.4 attacked this directly: shared credentials and a preference for joining the existing network rather than founding a new one. If you’ve read my Thread and OpenThread piece, that’s the spec’s front door finally getting a shared key hook. The 2026 reality is better — new hubs cooperate more often than not — but “better” is doing some work in that sentence, and plenty of homes still run accidental parallel meshes they don’t know about.
The symptom is always the same: a sensor that’s “unreachable” in one app and fine in another, because it joined the mesh belonging to the wrong ecosystem’s hub. Nothing in the home tells you this happened. The user experience of a partitioned mesh is indistinguishable from flaky hardware — which is exactly why the problem survived so long. Users blamed the devices; the devices were fine.
When the Cloud Dies
There’s a second axis of interoperability that nobody puts on the box: interoperability with the future.
Insteon, April 2022: servers switched off overnight, no warning, no farewell email. Tens of thousands of working devices went inert mid-week (a community later resurrected the service — the devices were fine; only the business had failed). Wink, 2020: a free hub abruptly demanded a monthly subscription, with a one-week ultimatum. SmartThings retired its first-generation hubs in 2021, and Best Buy’s Insignia line simply had its smarts amputated in 2019.
Different stories, one lesson: a device that can only speak to its maker’s cloud has the lifespan of its maker’s business model. That is an interoperability failure as real as any protocol mismatch — the device can’t talk to anything else when the only thing it ever talked to disappears.
And notice who pays. The company that shuts a cloud writes off a product line; the customer writes off a house full of hardware, the weekend spent installing it, and the automations built on top. The costs of closed design land years later, on someone else — which goes a long way to explaining why closed design keeps being chosen.
The Local-First Rebellion
While the giants negotiated standards, the community routed around them. Home Assistant grew from an enthusiast project into the most-installed open-source home automation platform on Earth, speaking thousands of integrations natively — Zigbee through a $30 dongle, Z-Wave, Matter, Thread, and a very long tail of reverse-engineered cloud APIs. Its Works with Home Assistant program now certifies what the community spent a decade proving: local control isn’t a hobbyist quirk, it’s a product requirement.
The rebellion’s influence runs beyond its install base. Vendors discovered that “works locally, no account required” shows up in reviews. Matter’s local-control architecture is, in spirit, the community position adopted by committee. And every dead cloud in the previous section pushed another wave of buyers toward the one platform whose business model can’t shut down their house.
Get Uladzislau Bayouski’s stories in your inbox
Join Medium for free to get updates from this writer.
It’s worth being precise about what the rebellion proved, because it’s not “open source wins”. It proved the interop problem was solvable with the devices already on the shelves. Home Assistant didn’t wait for vendors to agree on a standard; it treated every proprietary API, every cloud endpoint, every quirky Zigbee implementation as a translation job and did the work. A volunteer community out-integrated three trillion-dollar companies for a decade. Whatever the gardens said about technical difficulty, the difficulty was never technical.
Where Things Stand in 2026
Credit where due: the floor is real now. Thousands of Matter-certified products exist; every major ecosystem speaks the core device types; a bulb bought today will almost certainly answer to whatever assistant you run, locally, without an extra hub account. The Thread credential mess is healing. Even regulation is leaning in — the EU’s Data Act, applicable since September 2025, gives users rights to their connected-device data and makes switching between services a legal expectation rather than a courtesy.
But the ceiling is as proprietary as ever. The interesting features — the camera AI, the vacuum maps, the energy dashboards — live above the standard, in apps and subscriptions, because that’s where the recurring revenue is. The gardens adopted a common gate and kept the walls. Honest verdict: we’ve standardized the verbs and privatized the adjectives.
Where does it go from here? Watch two pressures. Regulation keeps tightening — data portability today, and it’s not hard to imagine functional-after-shutdown requirements tomorrow. And the resale argument is starting to bite: estate agents and landlords have discovered that a house of orphaned smart gear is a liability, not a feature. Both pressures push the same direction the buyers already are.
The Honest Diagnosis
Which points at the uncomfortable conclusion: interoperability stopped being an engineering problem years ago.
The protocols exist. The bridges exist. A committee of fierce competitors managed to agree a shared data model, which proves the technical fight was winnable all along. What remains is incentives. Lock-in measurably reduces churn. Differentiation lives in the proprietary layer, so the proprietary layer will always be one step ahead of the spec. Certification fees favor incumbents. An open device is easier to leave, and companies don’t optimize for being easy to leave.
The mental model that actually works: interoperability hell is a business-model problem wearing a protocol costume. Read every “works with” badge, every missing feature, every dead cloud through that lens and the smart home suddenly makes perfect, cynical sense.
For Builders and Buyers
Builders: ship Matter as your floor, not your ceiling. Core control through the standard, extras through a documented local API as well as your app. The vendors earning long-term trust in 2026 are the ones whose devices remain useful without their servers.
Don’t require an account for local function. A plug that needs a login to switch on is a rental pretending to be a product — and reviewers have learned to say so. Cloud features can be additive; they should never be load-bearing for a light switch.
Test multi-admin like you mean it. Pair your device to two ecosystems at once and use both for a week. That’s how your customers live, and it’s where half the “works with Matter” complaints are born.
Write down your end-of-life story. What happens to devices if the cloud closes? Firmware unlock, local fallback, escrow — Insteon’s ghost asks every one of your customers this question on your behalf.
Buyers: buy the standard, not the promise. Check for the Matter badge and check your category is actually covered — a certified plug is routine; a certified camera still deserves a second look at the feature list. And read one review from someone who runs a different ecosystem than the box advertises; that’s where the gaps show.
Run the airplane test. Cut the internet and see what still works. Thread, Zigbee and local-Matter devices pass. Cloud-Wi-Fi devices fail, and every one of them is a bet on its maker’s payroll.
Put fixtures on open radios. Anything wired into walls or ceilings — switches, thermostats, locks — deserves Thread or Zigbee and a local hub, because you’ll still own the house long after any app is discontinued. If you’ve read my DIY Smart Home Hacks piece, this is that instinct, formalized.
💡 Final Thought
The maze wasn’t malice. Every wall was a reasonable decision — a startup shipping fast on Wi-Fi, a platform protecting its users, a vendor keeping its best features for its own app. Interoperability hell is just what a thousand locally-rational choices look like from inside one house.
The way out isn’t another protocol. The language problem is solved; Matter proved the committee could agree on words. What’s left is a market problem, and markets listen to exactly one signal. Every purchase of a local-first, standards-floor device tells the industry the walls aren’t worth building; every cloud-locked impulse buy says the opposite.
Your devices already know how to talk. Buy from the companies that let them. 🧩
If you found this article insightful, follow for more deep dives into IoT, the smart home, and the tech shaping our connected world! 🔬
IMAGE PROMPTS
Four visuals: the walled-gardens topology and the border-router tangle genuinely need pictures; Matter’s slow rollout is clearest as a version chart; the dead-clouds timeline beats a paragraph of dates. The diagnosis and buying advice stay in prose.
Hero Image
Wide editorial banner image, 1200x630, on a clean off-white background. Conceptual illustration of three separate walled gardens drawn as rounded island plots, each enclosed by a low wall and containing a small cluster of smart-home device icons (bulb, speaker, camera in one; plug, thermostat, lock in another; vacuum, sensor, display in the third). Between the islands, a partly built bridge of puzzle-piece segments spans the gaps — some pieces in place, one piece visibly missing near the centre. A few small device icons queue at the walls’ gates. Minimal, sophisticated, editorial tech-illustration style. STRICTLY NO TEXT of any kind: no words, letters or numbers anywhere in the image. Muted teal, slate, and soft amber accent palette (amber for the puzzle bridge). Clean negative space. Professional, not sci-fi cheesy.
Visual 1 — Walled Gardens Map
Clean topology diagram on white background titled “Three gardens, one house” in dark sans-serif font. Three large rounded enclosures arranged in a triangle, labelled “Alexa”, “Google Home” and “Apple Home”, each containing three or four small device icons and a small gate icon on its wall labelled “Works with”. In the centre between them, one house icon labelled “Your home” with dashed lines running to each gate, and a person icon beside it labelled “You — the bridge”. Flat design, rounded shapes, slate enclosures with teal device icons and amber gates, sans-serif font. Clean minimal professional style.
Caption: “The ecosystems interconnect through exactly one component: you”
Visual 2 — Matter Coverage Progress
Clean infographic on white background titled “What Matter actually covers, version by version” in dark sans-serif font. A horizontal timeline with five stops, each a rounded card above or below the line: “1.0–2022” with “Bulbs, plugs, locks, thermostats, sensors”; “1.2–2023” with “Robot vacuums, appliances”; “1.3–2024” with “Energy, EV chargers, water heaters”; “1.4–2024” with “Batteries, solar, heat pumps”; “1.5–2025” with “Cameras — at last” (this card highlighted with an amber border). Flat design, teal cards on a slate timeline, sans-serif font. Clean minimal professional style.
Caption: “The most-requested categories took three years to reach the spec”
Visual 3 — The Border Router Tangle
Clean topology diagram on white background titled “Three border routers, three islands” in dark sans-serif font. LEFT: the outline of one house containing three hub icons labelled “Apple TV”, “Nest Hub” and “Echo”, each surrounded by its own dashed bubble holding four small Thread device dots; the three bubbles do not touch, and thin red “no connection” gaps separate them. RIGHT: a smaller “after” panel labelled “Shared credentials — one mesh”: the same house with a single dashed bubble containing all twelve device dots and the three hubs together. An arrow from left panel to right panel labelled “what should happen”. Flat design, teal bubbles, slate hubs, coral separation marks, sans-serif font. Clean minimal professional style.
Caption: “Border-router fragmentation — one home, three Thread islands”
Visual 4 — Dead Clouds Timeline
Clean infographic on white background titled “When the cloud dies, the device dies” in dark sans-serif font. A horizontal timeline from 2019 to 2022 with four marked events, labels alternating above and below the line: 2019 “Insignia — smart features switched off” (grey marker); 2020 “Wink — surprise subscription ultimatum” (amber marker); 2021 “SmartThings — first-gen hubs retired” (amber marker); 2022 “Insteon — servers off overnight” (red marker, slightly larger). Each label carries a small unplugged-cable icon. Flat design, sans-serif font, slate line with grey, amber and red markers. Clean minimal professional style.
Caption: “Cloud-only devices inherit the lifespan of their maker’s business model”
