For 15 years, hardware companies competed by making devices “smarter.” The rice cooker became the smart rice cooker. The speaker became the smart speaker. The TV became the smart TV. Product managers played guessing games with firmware, trying to hard-code every possible user scenario to improve the user experience. AI ends this. The software logic layer is becoming a universal commodity, turning devices into “physical primitives.” The future isn’t about smarter baked-in logic — it’s about building the best unopinionated physical shell for AI to inhabit.

The Smart Device Paradox

The rice cooker on your counter has three buttons: Rice, Soup, Porridge.

The heating element inside it can hold any temperature between 30°C and 180°C with precision — enough to make yogurt, sous vide a steak, or proof bread dough. But it won’t. A product manager decided, before the thing left the factory, that you would only ever need three modes.

Your smart speaker is the same story. It ships with hard-coded support for two streaming services (the ones that signed a business deal), a fixed alarm routine, and a morning briefing that goes weather → news → calendar, every single day, whether you care about the weather or not. Amazon poured billions into Alexa, hired 10,000 engineers to build Skills, and still ended up with a $200 egg timer.

The hardware is perfectly capable. It’s the logic layer — the part that decides what the device does — that was written by someone who has never been inside your house.

The Geek Escape Hatch

If you’re technical enough, you’ve always had an out. Buy a Raspberry Pi or an Arduino, wire up some sensors, and write your own automation logic.

The problem is most people can’t do this. And even if you can, the time you spend writing and debugging code probably exceeds the time you save. At best it becomes just another gadget you show off to friends while your wife rolls her eyes. (Don’t ask me how I know.)

Enter AI: The Universal Logic Layer

The AI doesn’t need you to set rules or tell it what to do. You leave the house — it knows, and the lights go off. No configuration, no voice command. It just understands the context.

You’re at home? The AI routes audio to your living room speaker. You step outside with your earbuds? It continues on your earbuds. No product manager foresaw this specific transition. The devices simply exposed their capabilities — “I can play audio, here is my location, here is my current status” — and the AI figured out the rest.

Here’s a real example from my own life. I was out and needed to reach my wife urgently. Her phone was dead and she didn’t notice my calls. My AI agent found the Sonos speaker in our living room, turned my message into speech, and played it loud enough to get her attention. It worked. No one at Sonos designed a “relay urgent spousal messages” feature. Sonos just built a damn good speaker. The AI didn’t use a feature — it used a tool.

And here’s the thing — AI models evolve weekly. When the logic layer lives in the cloud, the hardware becomes decoupled from the innovation cycle entirely. A three-year-old microphone array can suddenly understand you better overnight, without a single firmware update. The hardware doesn’t change; the intelligence keeps getting better.

The Dissolving Role of the Product Manager

Think about what a smart speaker PM does today. They spend months debating how many alarm tones to ship with (they’ll guess wrong), what the morning briefing order should be (they’ll annoy everyone), and whether to add “sleep sounds” in this firmware update or the next (the speaker can already play any audio right now).

In an AI-native world, the PM doesn’t need to make most of these calls. The speaker just needs to be a great speaker — clear audio, good microphone, reliable connectivity. The AI handles everything else. Not because someone programmed “morning routine v2.3” into the firmware, but because the AI understands you.

The PM’s job shifts from “design the experience” to “build the best physical tool.” The layer of opinionated software decisions that used to define a product gets thin. Vanishingly thin.

The Hardware Renaissance

This leads to a conclusion that feels backwards.

We’ve all heard “software is eating the world” beaten to death by now. But when the software logic layer becomes a commodity — when every device can plug into the same AI and get the same intelligence — the only remaining differentiator is the hardware itself.

When intelligence is commoditized, physical execution becomes the only moat.

Suddenly it matters who has the better microphone array, the lower-power chip, the camera sensor that works in near-darkness. It matters who makes the thing that feels good in your hand and survives being dropped.

Hardware is getting thinner — not physically, but architecturally. The layers of logic, opinion, and product-manager taste that used to sit between the raw capability and the user are dissolving. What’s left is the hardware itself, stripped bare, doing its physical job as well as it possibly can.

The rice cooker can finally make yogurt.