Everyone's chasing voice control. Your hand already knows where the switch is.
- Wayne Du Bruyn
- 2 days ago
- 4 min read

Every automation platform launch this year has pitched the same future: talk to your house, and it listens. Alexa+ remembers your preferences across sessions. Google folded its camera AI into a paid tier. The entire industry is racing toward voice and AI as the interface, as if pressing a button is something we're all meant to grow out of.
KNX Association published a piece in May making the opposite case, and I think they're right. Tactile buttons aren't a legacy holdover waiting to be replaced. They do something voice control structurally cannot.
The problem voice control can't design its way out of
A study presented at this year's CHI conference tracked thirty households using voice assistants over three years. The friction it documented wasn't a bug that better AI fixes with the next model update. Devices that reliably recognise one partner's voice and not the other's create a specific, recurring tension in a household, and that tension doesn't fade with use. It reshapes how much people trust the system, quietly, over months.
That's the part the industry keeps missing. Voice interfaces ask you to remember the right phrase, say it clearly, and then wait to find out whether the house understood you. Every one of those steps is a place for doubt to creep in. A button has none of them. You press it, you feel it give, you know. KNX's own framing calls this haptic certainty, and it's a good name for something real: physical feedback tells you your command registered before anything else has to confirm it.
There's research outside the smart home world that backs this up. A comparative study of touchscreens and physical buttons in car interiors found that drivers trusted a physical control without needing to look at it, where a touchscreen icon needed a glance to confirm it had actually worked. None of that is really about cars. It's about what a hand already knows how to do without your eyes getting involved.
Where this actually matters in a house
Think about the moments a smart home has to handle without becoming a production. Getting up at 2am and needing the passage light at 10%, not full brightness, not a voice command that risks waking anyone else. Adjusting a scene mid dinner party without narrating it to the room. A guest who has never met your voice assistant and doesn't want to learn its phrasing, but can press a labelled button with total confidence.
None of those are edge cases. They're most of what a house is actually used for, day to day.
A KNX installation handles this by design, not as a fallback. A keypad press is a direct command on the bus. No cloud round trip, no wake word, no chance the system misheard you and did something else entirely. Scenes get built once at commissioning and then just work, the same way, every time, for as long as the building stands. That reliability is also why it holds up here specifically: when City Power drops the connection along with the lights, a wired KNX keypad doesn't care. It was never waiting on the internet to begin with.
The aesthetic argument is not a small one
There's a design point worth taking seriously too. A well-made physical interface, whether it's brushed metal, glass, or a discreet flush panel, is something an architect can actually design with. It sits in the wall as a considered object, not a plastic puck that has to be placed somewhere and then apologised for. That's a real difference for anyone specifying a high-end interior, and it's part of why keypad placement belongs in the same conversation as the light fittings and the door hardware, not bolted on afterward.
Voice and buttons aren't actually competing
None of this is an argument against voice control. Ask the house to dim the lights while your hands are full of groceries and voice is genuinely the better tool. The point is that a smart home stops being reassuring the moment it only offers one way in, and the industry's current obsession with voice and AI-first interfaces is quietly narrowing that down to one option by default.
The brief for a real KNX installation keeps both. Voice and app control where they're convenient, a tactile keypad wherever certainty matters more than convenience, in the nursery, at the front door, beside the bed. Getting that mix right is a design decision, not an afterthought, and it needs to happen at first fix, while the wiring is still going in.
If you're planning a build and want an interface strategy that actually accounts for how people use a house, not just how a product demo looks, get in touch. Just a conversation about what the project needs.
Wayne | KNX Logic
wayne@knxlogic.co.za | 082 564 3982 | www.knxlogic.co.za
Sources: The psychology of control: Why tactile buttons still matter in a voice-first world (KNX Association, 12 May 2026, knx.org); Finding a Home for Voice Assistants: A Domestication Calculus Across Three Years and Thirty Households (CHI 2026, ACM Digital Library); The influence of tactile feedback in in-vehicle central control interfaces on driver emotions (ScienceDirect, 2024).




Comments