jj jashjacob.com
← field notes smart home 1 May 2026 4 min read

Why I Chose Aqara for My Home

I spent months talking myself into Zigbee, then bought three Wi-Fi devices on purpose. How I ended up on Aqara for a hundred-device house, and where the clean story leaks.

I spent months talking myself into Zigbee, and then I bought three Wi-Fi devices.

On purpose. The FP2, the G100 and the Magic Switch. Three decent sized holes in the clean mesh-networking story I like to tell, all put there deliberately, and they turned out to be some of the most useful devices in the house. The first version of this post kept that confession near the bottom. It’s the actual story, so this time it goes first.

The reason it took months to reach a decision I’d end up bending: most smart home advice is written for someone who wants four bulbs and a speaker. If that’s you, honestly, buy whatever’s on sale. It’ll be fine. I was planning for something closer to a hundred devices, and that’s not the same problem scaled up, it’s a different problem entirely. Past ten or fifteen things you stop caring about features and start caring about really boring stuff. How the devices talk to each other. What happens when the internet drops. Whether the batteries are going to have you up a ladder every six months.

The four rules

A smart home is infrastructure, not gadgets. Which sounds like a line, but it decides what you buy. Gadget thinking buys the trick you want today. Infrastructure thinking asks whether the thing still works in four years when there are ninety more of them, and whether it keeps working when the company that made it has a bad quarter.

So the list was short, and in order:

  1. Local control. The house doesn’t stop working when the internet does.
  2. Zigbee for anything battery powered. Wi-Fi devices eat batteries and clog your airtime.
  3. Sensors cheap and small enough to put everywhere instead of rationing them.
  4. It has to disappear into the house. I didn’t want to live in a server room.
the rule

“Zigbee unless there’s a real reason” is a useful rule. No rule at all is how you end up with fourteen apps.

That list is what landed me on Aqara.

What I actually run

devicejobradio
Hub M3the middle of everything · IR blaster gets the dumb AC inzigbee coordinator · PoE
FP2 presenceknows the room is occupied, not just that something movedwi-fi
20A switcheswater heater and ACs · power monitoring on the loads that cost moneyzigbee
T2 relay modulesbehind the wall switches, which still work like wall switcheszigbee
A100 lockfront door · fingerprint and Home Keyzigbee
G4 doorbellthe doorbell
G100 cameracamerawi-fi
Magic Switch480×480 touchscreen where a 3-gang switch used to bewi-fi
Amber rows are the exceptions, and the story

Two of these deserve more than a row.

The Hub M3 is the middle of everything. Zigbee coordinator, Thread border router, Matter, and it takes PoE, so it sits on a network drop instead of a wall wart. It’s also got an IR blaster in it, which sounds like a gimmick until you realise it’s how a dumb AC gets into your automations without replacing the AC.

And the FP2, which is the one I’d buy again first. It’s mmWave radar rather than a PIR motion sensor, and the difference is the whole point. A PIR sees movement, so it turns the lights off on you while you’re sitting still reading. The FP2 sees that a person is there, and it can split one room into zones and track a few people at once. It’s brilliant. It’s the single device that changed what I thought automation could be, and it’s Wi-Fi, and I don’t care.

Auditing my own rules

Four rules going in. Here’s what happened to each of them, now that I live inside the result.

ruleverdictwhere it leaked
Local controlmostly heldthree Wi-Fi devices lean on Aqara more than the mesh does
Zigbee firstbent, on purposeFP2 · G100 · Magic Switch
Sensors everywhereheld in theoryI under-bought them early, which was me, not Aqara
Disappears into the househeldrelays hidden, switches unchanged
The leaks were all choices, which is the only kind I can defend

So the architecture isn’t pure. I optimised for Zigbee and then made an exception every time the exception was worth it, and the exceptions turned out to be some of the best devices in the house. The bill arrived later: Wi-Fi coverage stopped being a convenience and became load bearing, which is a problem I got to discover separately, and wrote up separately too.

Was it the right call

Yes, and I’d do it again. Not because Aqara is perfect. The app is trying to do too much, and there are more overlapping hub generations than there should be. But the sensor range is genuinely good, the hardware disappears into the house, and the things that should be local are local.

What I’d tell someone planning the same build

  • Buy sensors, not switches. Switches are remote controls with extra steps. Sensors are what makes the house do things on its own. I under-bought sensors early and it’s the one thing I’d properly change.
  • Pick your protocol before your first purchase, then allow yourself exceptions. The rule does the work even when you break it.
  • Presence beats motion. If you buy one thing to see what modern automation actually feels like, make it an mmWave presence sensor, not a PIR.
  • Check the boring specs. The 20A rating and the PoE port aren’t exciting, and they’re the two decisions I’m still happiest about.

The clean story

Every setup post wants to tell you a clean story, and mine has three Wi-Fi shaped holes in it. I put them there, and the house works better for them. The only version of this decision I’d regret is the one where I pretended otherwise. If you do this, you’ll make the same compromises.

Make them knowingly instead of pretending the architecture is pure.

Measured, not assumed

4
criteria applied
8
aqara device types
100
% local execution
0
subscriptions