Whole-home smart control has three layers: the host senses and decides, the sensors collect data, and the control panel handles the human interface. Because decision logic sits on the local machine, HVAC and security keep running their basic functions when the network drops; cloud-dependent systems lose most of their automatic control once offline.
The host is the decision center: it receives sensor data, judges the state, and issues commands — when CO₂ passes its threshold, it tells the fresh-air unit to raise airflow. The control panel is the human interface: it shows status, takes your input, and passes commands to the host. It stores no decision logic itself. Selling the screen as the brain and demoting the host to a gateway means one broken panel paralyzes the whole house.
This is the most commonly confused point, and the one where buyers are most easily misled.
The hostis the decision center. It receives sensor data, judges the current state, and issues control commands. If CO₂ in a bedroom passes the threshold, the host decides to raise the fresh-air volume and commands the fresh-air unit to do it. That chain runs inside the host.
The control panelis the human interface. It shows status, takes your input, and forwards commands to the host. It stores no decision logic and makes no judgments of its own.
Many systems sell the control panel as the "brain" and reduce the host to a gateway. The result: when the screen fails, the whole house's logic collapses; offline, the screen cannot reach the cloud and many scenes stop working.
It depends on the architecture, not the brand. In a local system, automatic temperature, humidity, and CO₂ regulation, fresh-air and dehumidification operation, local security alerts, and panel-to-host operation on the same local network all stay available; remote viewing and control, remote alert push notifications, and most voice assistants do not. Precisely put: basic control and local alerts do not depend on the network; remote notification does.
The answer depends on the architecture, not on the brand. Our approach splits functions into two layers:
| Function | Offline | Notes |
|---|---|---|
| Automatic temperature, humidity, and CO₂ | Available | Logic is local; sensors and actuators use a local bus |
| Fresh-air and dehumidification | Available | The HVAC loop is independent of the network |
| Security alerts (local) | Available | Local sound and light alerts, local logging |
| Remote security alerts | Unavailable | Needs a network to reach your phone |
| Remote viewing and control | Unavailable | Network required |
| Voice assistants | Mostly unavailable | Speech recognition usually runs in the cloud |
| Panel operation on the local network | Available | The panel and host share one local network |
The table above describes functional boundaries at the architectural level. Actual per-project behavior follows the configuration confirmed after commissioning; a functional boundary list is provided at handover.
It detects presence and movement, trends in heart and breathing rate at rest, and posture patterns consistent with a suspected fall. It cannot diagnose illness, cannot replace medical equipment, and cannot guarantee detection of every fall (carpets, furniture blocking the sensor, and several people present at once all reduce accuracy). Its role is assistive alerting; it does not call emergency services.
What it can detect: presence and movement, trends in heart and breathing rate at rest, and posture patterns consistent with a suspected fall.
What it cannot detect: it cannot diagnose any illness, cannot replace medical monitoring equipment, and cannot guarantee 100% detection of every fall — carpets, furniture blocking the sensor, and several people present at once all reduce accuracy.
So we position it asassistive alerting: when a suspected fall is detected, an alert goes to family members, who decide whether further action is needed. It does not call emergency services, make medical judgments, or take responsibility for emergency response.
For new builds and full renovations, whole-home fiber lasts longer: higher bandwidth ceiling, immunity to electromagnetic interference, longer runs, longer service life. The trade-offs are optoelectronic conversion at the endpoints, higher cost, and more demanding installation. In a renovation with existing cable that works normally, there is no need to force a switch to fiber — fiber as the backbone and cable at the endpoints balances cost against headroom.
For new builds and full renovations, whole-home fiber is the longer-lasting choice: higher bandwidth ceiling, immunity to electromagnetic interference, longer runs, longer service life. The trade-offs are optoelectronic conversion at endpoint devices, higher cost than cable, and more exacting installation work (bend radius, splice quality).
In arenovation project, if the existing cable works normally there is no need to force a switch to fiber. Fiber as the backbone and cable at the endpoints balances cost against headroom.
Sensor data and operating records are stored on the local machine by default. When remote features are on, the necessary status data is encrypted and transmitted to the cloud for viewing on a phone. Voice audio follows the rules of the voice service you choose. At the design stage we list explicitly which data stays local, which is uploaded, what it is used for, and how to turn it off.
Sensor data and operating records are stored on the local machine by default. With remote features on, the necessary status data is encrypted and sent to the cloud for viewing on a phone. Audio from voice interaction is handled under the rules of the voice service you select.
At the design stage we list explicitly which data stays local, which is uploaded, what the upload is used for, and how to disable remote functions. That is both a privacy compliance requirement and something a homeowner should know.Privacy Policyhas the full details.
Security should be layered: perimeter and entry (locks, door and window contacts, video intercom) answer whether anyone came in; the interior (presence detection, millimeter-wave radar, smoke and gas) answers what is happening inside; equipment safety (leak shut-off, distribution box temperature, smart panel faults) answers whether the system itself is sound. The third layer is the easiest to overlook and the one most often triggered.
Security is not a pile of sensors. It is layered: perimeter, entry, interior, and equipment safety.
The third layer is the easiest to overlook, yet in long-term living it is the one most often triggered. Leak shut-off and distribution fault warnings are imperceptible day to day, and when something goes wrong they can save a major repair.
Stability depends on three things: the ratio of wired to wireless, the number of devices, and the quality of commissioning. Wired circuits are markedly more stable than fully wireless ones. More devices and more complex scenes mean a higher chance of failure, so we limit the number of scenes and keep only those used frequently.
Third-party devices that support common protocols can be integrated, but cross-brand automation is less stable than within one ecosystem. At the design stage we indicate where devices from the same ecosystem are recommended.
Local control logic runs for years without upgrades. Firmware updates carrying security patches or feature changes should be applied when prompted, but installed while someone is on site, avoiding the risks of silent remote upgrades.
Core functions should require no operation at all — the system adjusts itself and older residents simply live as usual. Anything that still needs operation is provided as a physical switch or by voice wherever possible. That is a design principle, not an add-on.
Indoor cameras are the devices that demand the most caution. Our default design installs no indoor cameras, using non-imaging sensors such as millimeter-wave radar instead. If a homeowner asks for cameras, data storage and viewing permissions are written into the contract.
To find out whether your house is a fit and how a scheme would work, call us or add us on WeChat. We answer one to one.