XingBan OS · built in-house

This is an operating-system layer, not an application platform.

XingBan OS is built in depth on the full OpenHarmony foundation and uses our own perceive–reason–act architecture to admit six elements — people, devices, materials, environment, systems and data — through one set of interfaces.

Foundation
Full OpenHarmony stack, extended in-house.
Architecture
Perceive → reason → act.
In-house
The operating-system layer is our own implementation, not a wrapper.
Runs on
XingBan Ark, or an institution's existing servers.
02

Architecture

From perception to governance, six layers with defined responsibilities.

Perception collects, data retains, cognition judges, orchestration assigns, application executes and governance draws the boundaries. All six share one interface.

L0 perception
Bedside, wearables, access control and radar, admitted through one interface.
L1 data
Stored on site, one definition, traceable.
L2 cognition
The model judges risk and need at this layer.
L3 orchestration
Breaks a judgement into tasks and assigns them to a post.
L4 application
Nine digital workers and six scenarios run here.
L5 governance
Permissions, records, audit and data boundaries.
03

One intake

Every element on site is admitted through one interface.

Site elements fall into six categories: where people are, whether devices are working, whether supplies are sufficient, whether the environment is safe, whether systems are connected and whether data is accurate. Only when all six are admitted to one layer does unified computation become possible.

People
Residents, staff, families, visitors.
Devices
Bedside terminals, wearables, access control, cameras, radar.
Materials
Medication, consumables, food, equipment.
Environment
Temperature, humidity, smoke, water, noise.
Systems
Existing HIS, finance, access control, CCTV.
Data
Records, notes, ledgers, metrics.
04

Three hard constraints

Data stays in the building and is still available offline.

Three product constraints: resident data is among the most sensitive categories there is; a night-time alert cannot wait for the network; and an institution's systems cannot stall because the external network wobbles.

Data stays in
Raw data remains on site; only conclusions are shared.
80% local
Target for inference completed on site.
Works offline
Critical judgements do not need the external network; sync resumes on recovery.

How it connects

Connection is a continuing programme, not a purchase.

We align with existing systems and data definitions first, then agree the scope. Equipment that can be reused is retained; what is missing is added unit by unit. No wholesale replacement is required.

Book a demo

Title

Thirty minutes to observe nine digital workers operating in a real care setting. An advisor will call to confirm the time and focus.