AIoT capability

What we do, and what we explicitly do not

The limits are stated plainly. For a system that must run for years, knowing what a supplier will not promise is worth more than a list of technical terms.

Shared foundation

The device-edge-cloud layers

What matters is not the number of layers but which logic sits at the edge and which on the platform. That decides what still works when the network drops.

  1. Field 01

    Field assets

    Equipment, control cabinets, meters, sensors and energy systems.

  2. Edge 02

    Edge onboarding

    SG / BEC gateways with SDS / IOC modules handle protocol onboarding and local control.

  3. Data 03

    Data governance

    Protocol onboarding, point modelling, data validation and edge buffering.

  4. Platform 04

    Platform

    AIoT operations platform, mobile apps, dashboards, alarms and work orders.

  5. Ops 05

    Continuous operations

    Trend analysis, anomaly detection, diagnostic suggestions, reporting and ongoing service.

Available now

What we can demonstrate today

Each item below is backed by running projects and can be verified during a pilot.

Unified acquisition and management

IoT-wide acquisition with one ledger, one model and one permission scheme for device, energy and environment data.

Real-time edge judgement

Edge-side judgement, local interlocks, offline buffering and event upload — no dependence on constant cloud connectivity.

Multi-dimensional anomaly detection

Thresholds, state, loss of contact, trend anomalies and correlation across multiple sources.

From alarm to closure

Alarm tiering, de-duplication, notification, work orders, inspection, reporting and knowledge capture.

Assisted diagnosis and reporting

Assisted diagnosis, health scoring, report generation and knowledge Q&A for service and operations staff.

Where AI sits

Edge AI decides now. Central AI improves over time.

One data foundation, two roles: the edge judges and acts in real time without depending on the cloud; the centre aggregates across sites to diagnose, forecast and accumulate policy.

Edge AI

Close to the field, judging in real time

  • Real-time detection Thresholds, trends and anomaly patterns.
  • Local interlocks Rule execution and on-site interlocking.
  • Reliable acquisition Offline buffering with backfill on recovery.

Central AI

Aggregating globally, optimising continuously

  • Global diagnosis Cross-device correlation and assisted root-cause analysis.
  • Trend forecasting Anticipating risk, efficiency and maintenance needs.
  • Policy capture Rules, models and knowledge accumulate over time.
  1. 01 Detect earlier Anomalies identified in real time.
  2. 02 Diagnose better More evidence behind each diagnosis.
  3. 03 Act faster Interlocks, notification and work orders in step.
  4. 04 Keep improving Experience keeps accumulating.

Roadmap

The five levels, and where we actually are

The point of levels is to keep "possible later" separate from "deliverable now".

  1. L1

    Visual monitoring

    See the state of equipment, environment, energy and people.

    Deliverable now
  2. L2

    Proactive alarms

    Notify proactively on thresholds, loss of contact, state and trend.

    Deliverable now
  3. L3

    Closed-loop operations

    Alarms reach people; dispatch, handling, feedback and statistics are traceable.

    Deliverable now
  4. L4

    Intelligent optimisation

    Suggest maintenance, efficiency and scheduling actions from historical data.

    Selected modules
  5. L5

    Bounded autonomous operations

    Execute and optimise automatically within explicit safety boundaries.

    Not a commitment

Explicit limits

What we do not promise

These are stated here to save time during evaluation and to avoid a gap between expectation and delivery.

  • Whether a device can be onboarded depends on the equipment list, protocol, point table and site conditions. We do not claim to connect anything.

  • Energy management is a scene inside packaged solutions. It is not a full EMS and does not perform real-time dispatch, production control or safety protection.

  • It does not replace inverters, PCS, BMS, charge controllers, field protection, fire interlocks or production control systems.

  • We do not promise unverified savings percentages, alarm accuracy, remote resolution rates, lead times or prices.

  • Customer data is not used for cross-customer model training by default. Ownership, access, tenant isolation and anonymisation are set in contract.

  • Where remote control is involved, we first define safety boundaries, permission tiers, operation auditing and the responsibility split with the original system.

Delivery path

From diagnosis to copy

A single-site, verifiable, repeatable pilot comes before any comprehensive platform. Samples are chargeable in principle, with credit possible against volume orders.

  1. 01

    Diagnosis

    Confirm equipment, protocols, sites, the pain, network and acceptance goals.

  2. 02

    Selection / sample

    Limit the devices and functions, then prove onboarding, alarms and pages. Samples are chargeable in principle, with credit possible against volume orders.

  3. 03

    Single-site pilot

    Produce the configuration list, rules, work orders, reports and service boundary as a repeatable acceptance sample.

  4. 04

    Volume copy

    Lock templates, an implementation handbook, pricing and the ongoing-operations mechanism, then copy to more devices or sites.

Core strengths

Fast to connect, dependable in operation, closes the loop

Three claims a customer can check against their own situation, each with the specific practice behind it.

Fast to connect

Protocol and device templates

  • Modbus, RS485/232, SNMP, MQTT, DI/DO and third-party APIs
  • Protocol templates and device models cut repeat development
  • Per-site deployment, protocol adaptation and acceptance scope are stated

Dependable in operation

Stability and data quality

  • Online, offline, state, threshold and trend anomalies managed in one place
  • Edge buffering, local interlocks and multi-target upload are configurable
  • Alarm tiering, de-duplication, escalation and ownership are configurable

Closes the loop

Handling and operational review

  • Notification, remote handling, dispatch, inspection and feedback all leave a trail
  • Produces monthly reports, fault statistics and service evidence
  • Faults and fixes accumulate into reusable experience

Deployment

Four delivery modes, with responsibility stated

SaaS cloud service

Quick to start, billed per device or site — suited to samples and standardised rollout.

On-premise / private

Deployed in the customer's own environment to meet data, network and access requirements.

OEM branding

Pages, tenants and product configuration delivered under the partner's brand.

API integration

Opens data and events to MES, ERP, property, energy or the customer's existing platform.

What every contract must state

  • Protocol and point scope
  • Data ownership and access
  • Remote control and security policy
  • Integrations and custom development
  • Cloud, traffic and notification costs
  • Implementation, acceptance and SLA

Project checklist

What every project must confirm

This list comes from project reviews. Anything left vague becomes a scope or acceptance dispute later, so we confirm each item during assessment.

Start the assessment
  • Customer type and responsibility split
  • Device and site counts
  • Protocols and point tables
  • Data rates and retention
  • Network and power conditions
  • Deployment: SaaS / on-premise / private
  • Platform functional boundary
  • Alarm and work-order scope
  • Remote control permissions and auditing
  • API and OEM requirements
  • Timeline and acceptance criteria
  • Support responsibility and SLA