The ESP32 is a wildly powerful little chip that’s near perfect for making your own smart home devices. If you’re using Home Assistant, ESPHome makes the job even simpler. You’d be surprised by just how much value an ESP32 and a few cheap sensors can add to your home.
But if you’re attempting to make everything smart in your home by yourself, you should probably give it another thought. Some smart home products are better bought than built, and knowing which ones to buy is sometimes the smarter choice.
Buy the smart bulb, keep the LED strip
Heat, enclosures, and mains don’t care how good your YAML is
Making an RGB LED glow the color you want with an ESP32 is perhaps the most basic thing you can do with the microcontroller, but it’s not the same engineering problem as building a dependable screw-in bulb that works with regular household power terminals and lamps. Sure, you can have an addressable RGB LED strip working with an ESP32 in an hour, but handling heat, mounting, and making an enclosure are separate issues that’ll eat away your time.
For standard house lighting, it’s almost always better to buy commercial smart bulbs, especially considering how low they can be in cost. You can easily find fully RGB smart bulbs for under $10 on Amazon, which is only slightly more than it’ll cost you to build one. Not to mention that the enclosure, mounting, and any safety hazards are handled for you. Your custom lighting ambitions are best served on separate, low-voltage projects rather than rebuilding something that screws directly into a mains powered socket.
Smart plugs aren’t a soldering-iron flex
Your homemade plug is one stuck relay away from a bad day
You can build yourself a smart plug with an ESP32 and a relay, but that’s nowhere close to a finished outlet. Even ESPHome’s flashing guide warns that an open device connected directly to mains can be an electrocution hazard.
The load matters too. Exceeding a plug’s supported ratings, or switching loads with high inrush current like motors and power supplies, can make relay contacts stick together, leaving it unable to disconnect the appliance. Considering you can get a 3-pack of smart plugs for roughly $25 on Amazon, it’s just not worth the time and risk. And similar to light bulbs, most commercial smart plugs are one integration away from being added to Home Assistant, so you can continue using your self-hosted platform without worries.
The reasoning is simple: a homemade box shouldn’t be the weakest link between the wall and everything you plug into it. Save your customization for the automation rules, not the mains wiring inside.
Smart toasters and the apology I owe mine
Anything with a heating element deserves a certification sticker
Smart home appliances have their own allure, which is exactly the reason you’ll find dozens of them on the market, no matter what you’re looking for. However, buying a smart appliance that’s been properly tested is one thing, and jury-rigging your own is another. As someone who’s tried to make his toaster smart and now owes it an apology, I can speak from experience that some devices are better left dumb.
That said, if you must have a smart appliance, it’s better to buy one that’s been appropriately safety-certified and tested, especially when it comes to anything with a heating element. Buying beats building here, but declining an unnecessary smart upgrade can beat both.
Don’t DIY the door you need to open
Locked out by your own weekend project is a special kind of bad
For a front door, I’d want something more than a motor turning a deadbolt on my workbench. Being unable to enter your house because your weekend smart home project has failed while you were out at work is not a great feeling, especially if you haven’t given yourself alternatives, something commercial locks do several times over.
All locks are different, meaning you’ll likely have to design and 3D print some sort of geared mechanism before you can even consider using a microcontroller to control your main door’s lock. If anything from the firmware you flash on the controller to a part in the overall assembly fails, you risk being locked out of your own home.
This is especially important if you live with other people who’ll need to access the main door regardless of your presence. You might be able to troubleshoot a non-functional lock from the outside and get past it, but not everyone would want to. If you must have a smart lock, it’s better to look at commercial options that provide stable connectivity, software, and backup ways of entering so you don’t get locked out of your house.
Smoke and CO alarms are not a hobby project
Your calibration choices shouldn’t decide who wakes up
As always, adding a few cheap sensors is all it takes for your ESP32 to detect and report any harmful gases around you. But a sensor reading on the dashboard isn’t evidence that your homemade alarm meets the several safety certifications commercial alarms have to pass before they can enter the market.
Buy an independently certified alarm appropriate to your region, with connected features as an addition to its protective function, not the reason to trust it. It’s best to prioritize an audible warning that works without your home automation server or internet connection. I’m happy to experiment with environmental data in my lab, but I wouldn’t make a sleeping household depend on my calibration choices, notification settings, or ability to anticipate every possible failure mode.
Build the fun stuff, buy the stuff that can hurt someone
Nobody agreed to be your beta tester
The question here isn’t whether you can build something. It’s whether building it is the best use of your time, and whether you’re comfortable dealing with any failures afterwards. My rule is to buy safety-critical hardware, then get creative with how it fits into everyday life.
There are still tons of smart devices you can make that’ll add value to your life and save you money. For example, making your own smart LED strips instead of buying something like a Philips Hue, is a much better way of saving money and building something that’ll work well. If you have to explain failures of a smart device to someone who never agreed to become your beta tester, you’re likely better off buying the device instead of building one.