The Dashboard My House Actually Needed
Every consumer family-dashboard product is either a walled garden or doesn't talk to the smart home you already run. So I built one that does.
We wanted one thing on the wall in the kitchen: chores, groceries, the calendar, and whatever was playing on Plex, all in one place, always current. Every product that claims to do this is either a subscription-gated app that doesn't integrate with the smart home infrastructure we already run, or a DIY dashboard framework that requires more YAML than a family actually wants to maintain. So I built Kith — a dashboard for the wall, wired directly into the automation systems already running the house.
Building for people who didn't ask for a beta
The hardest part of building Kith wasn't the dashboard. It was that the users are my own family, on a screen they look at every single day, and "good enough for a demo" doesn't survive contact with someone checking the grocery list before running to the store. Bugs that would never surface in a controlled test show up immediately when real people use something daily under real conditions.
The goals screen is the clearest example. A scoping bug in how the tab controls tracked state meant goal progress could silently apply to the wrong category if you switched tabs at the wrong moment — invisible in any demo, obvious the first time someone actually used it to track something they cared about. That's the tax of building software you live with instead of software you ship and walk away from: the bugs that matter are the ones only real usage finds.
Making the smart home actually feel real-time
A big piece of Kith is presence and status from Homey — who's home, what's running, what changed. The first version polled for updates, which meant a five-to-fifteen-second lag between something happening and the dashboard reflecting it. That's an eternity on a screen meant to feel like a live status board, not a page you refresh.
Switching to a real-time connection over socket.io closed that gap to near-instant, with polling kept only as a fallback if the live connection drops. It's a small technical change with an outsized effect on whether the dashboard feels alive or feels like a cached snapshot — and it's the difference between something people glance at and trust, versus something they learn to double-check.
Voice, without the walled garden
We wanted "set a timer" to work from across the room without buying into a closed ecosystem. That meant building a custom conversation-triggered automation tied to a dedicated timer entity, surfaced back on the Kith display — rather than relying on a commercial assistant's private timer state that the dashboard has no way to read.
What building for your own house teaches you
Every feature in Kith exists because something about managing a household with off-the-shelf tools was actively annoying, not because a market study said families wanted it. That's a smaller and stranger way to find product ideas than most product processes recommend — but it's also the most honest signal I've found. If a friction is real enough to fix for one household, it's usually real enough that other households feel it too, even if they haven't said so yet.