In the past ten years of making smart home products, we have witnessed too many so-called “pseudo-smart” phenomena.
The user shouts “turn off the lights” three times at the speaker, and it slowly executes it; the sweeping robot is stuck under the sofa, and a pop-up window on your phone asks you to “manually rescue”; the air conditioner is turned on at 22 degrees, and you are shivering from the cold, but it thinks everything is normal.
These scenes are familiar to everyone.
In the past few years, the industry has equated “voice control” with “intelligence” and packaged “App remote control” as “whole-house intelligence.”
Consumers paid for it at first, but after the novelty wore off, they began to complain.
Today, the industry has reached an inflection point.
In the second half of 2024, visits to multiple smart home exhibitions and brand conferences found that almost all leading players are downplaying “voice interaction” and instead focusing on “AI
Agent”, “active intelligence” and “scenario self-decision making”.
This is not a simple change of marketing rhetoric, but a migration of the underlying technology paradigm.
The voice control experience has a low ceiling. It is essentially a one-way logic of “people commanding machines”. The burden of intelligence is entirely on the user, which is no different than moving the switch from the wall to the mouth.
The maturity of AI large model technology has opened up a new path for smart homes, allowing devices to learn to understand people’s intentions and even make decisions for people.
To understand this change, the AI capabilities of smart homes can be divided into three levels.
The first level is to listen to instructions. Currently, most products are in this state. The voice assistant receives clear instructions and executes them, but the boundaries of capabilities are clear. Users need to predict needs and translate them into machine instructions, which results in high cognitive load.
The second layer is to understand intent, which is the key direction of the industry in 2024-2025. AI combines context to understand the real needs of users, changing from “command-driven” to “intent-driven”. Users only need to express vague needs, and AI is responsible for “translation” and execution.
The third level is active service, which is the final direction.
For example, AI learns user patterns, turns on air conditioners, water heaters, etc. in advance, and can also adjust services according to abnormal situations.
At its core, users don’t need to speak at all.
From the data point of view, China’s smart home market will exceed 800 billion yuan in 2024, a year-on-year increase of about 15%. The market has entered a parallel stage of stock replacement and experience upgrade.
IDC predicts that global smart home device shipments will exceed 600 million units in 2025, and the proportion of devices equipped with AI capabilities will increase significantly.
The implementation of the Matter 1.3 protocol has solved the “ecological island” problem and provided a data base for AI global decision-making.
Each core player also has a different AI layout.
Xiaomi has launched an upgraded version of “Xiao Ai” equipped with a self-developed large-scale model. It uses Xiaomi’s Pengpai OS as the base to connect mobile phones, cars, and home devices, and combines external information to make more accurate decisions.
Huawei adopts a “whole-house intelligent host” architecture. The whole-house intelligence 5.0 released in 2024 has “AI spatial perception” and realizes a “perception-decision-execution” closed loop.
Haier has embedded the capabilities of home appliances into “home scenarios” and upgraded from “device intelligence” to “life service intelligence”.
From the perspective of a product manager, this AI change has put forward new capability requirements.
It is necessary to shift from “functional design” to “scenario design” and focus on people’s complete experience; from “deterministic interaction” to “fuzzy interaction” and have the ability to “fault-tolerant design”; from “device manager” to “system manager” to understand the collaborative logic of the entire ecosystem.
AI does not make devices smarter, but makes the home warmer.
A good smart home should be like caring family members. From “voice control” to “active service”, it is a paradigm shift in the industry from “tool thinking” to “service thinking”.
As an industry practitioner, I feel that we are getting closer to the “real smart home”.
