Blog / smart home
Is Home Assistant Worth It for a Normal House?
Local control fixes the cloud outage problem, but it comes with a maintenance tail. Here is where a hub earns its keep and where it does not.
On August 11, 2026, PetLibro's cloud went down and a lot of cats and dogs did not get breakfast.
The interesting part is not the outage. Every service goes down eventually. The interesting part is that the company's position was that feeding schedules live on the feeder itself and should keep running with no internet, while owners were posting that their feeders sat there doing nothing. The CEO ended up saying publicly that they were looking into reports of devices not operating as intended.
"Works offline" is a claim. Until you have unplugged your modem and watched the thing run anyway, that is all it is. I wrote recently about what happens when the vendor behind a device stops paying the cloud bill entirely. This is the same problem on a shorter timescale, and it points at the same answer.
What a local hub actually changes
This is the argument for running something like Home Assistant. It is an open source home automation platform owned by the Open Home Foundation, a Swiss non-profit, and it is built around local control: the automations run on a small computer in your house rather than in somebody's data centre. Your schedule fires because a box on your shelf decided it was time, not because a server in Virginia sent permission.
That is a real difference in behaviour, not a philosophical one. When your internet drops, the lights still respond and the schedules still run. You lose remote access from outside the house, which is the correct thing to lose.
It also means the vendor cannot take features away from you in a firmware update, and cannot start charging monthly for something that used to be included.
The part people find surprising: reflashing
A lot of inexpensive smart plugs, switches, sensors and LED controllers are the same handful of ESP microcontrollers inside a different plastic shell. ESPHome, another Open Home Foundation project, lets you write a short config file that becomes the firmware for those devices. You replace what the manufacturer put on there with your own build, and the device stops talking to their cloud entirely. It talks to your hub, on your network, and nowhere else.
You often end up with more than you had. Vendors hide things: a sensor that reports to their app once a minute is often capable of far more, along with readings the app never showed you. The account and the rate limits go away, and the device keeps working long after the company loses interest in it.
The honest caveats. Not everything can be flashed, the ones that can often need the case opened, and some manufacturers have moved to chips the open firmware does not support, sometimes specifically to stop this. You are also voiding the warranty. The hands-on version belongs in a walkthrough rather than here, so I will leave that for the Friday series.
The cost nobody puts in the pitch
A local hub is a system you now own. It updates. Updates occasionally break an integration you cared about. Someone in the household has to be able to fix it.
That last one is the whole decision, so I will put it plainly: if the answer to "who repairs this when I am away for two weeks" is nobody, do not build the house around it. A smart home that only one person understands is a single point of failure wearing a nice dashboard.
For most families the right shape is a middle path. Buy devices that work locally out of the box, which increasingly means Matter and Thread, Zigbee gear paired to a hub you own, and cameras that record to something in your basement rather than only to a subscription. Keep the vendor apps for the things where they are fine. Add a hub when you actually want the hobby, or when the stakes justify the effort.
Lighting is the easiest place to see the difference. Lutron is the usual example: the hub runs on ethernet rather than Wi-Fi and keeps working when the internet is down, and the dimmer on the wall is still a dimmer no matter what the network is doing. My company installs it, so treat that as a bias I am stating rather than hiding, and the installation side of it is written up over at RelayPoint. The principle stands with or without me: control that lives in the wall beats control that lives in an account.
And whatever you do, keep a dumb fallback for anything with a real consequence: heat, water, locks, and yes, food. A bowl works during an outage. I have made the same argument about which systems need to survive a power cut, and the logic is identical when the thing that failed is a server rather than the grid.
For a small business, raise the bar
Everything above still applies, with less room to shrug. A hobby stack running a retail floor or a rental property is fine right up until the person who built it changes jobs. If you are going to run local automation in a business, someone has to own it in writing, it needs to be documented, and it needs a manual override that a part time employee can operate without a phone.
That is not an argument against local control. Local control is usually the more resilient choice. It is an argument for treating it like infrastructure instead of a weekend project, which is a version of the point I make about gear you already own and should not rip out.
If you want a second opinion on a smart home that has grown past the point of anyone understanding it, that is an assessment and I quote it as a fixed fee. If the fix involves wiring, cameras or getting equipment properly installed, my company RelayPoint Technologies does that work across Southern Ontario. Otherwise, tell me what broke and I will tell you whether it is worth rebuilding or worth leaving alone.