FAQ
The questions we hear most during evaluation
Direct answers to the questions that come up during evaluation, including what we cannot promise. If yours is missing, just ask — it is faster than reading a proposal deck.
01
About Tongwei
What does Tongwei Smart Technology do?
Tongwei provides industrial-grade AIoT operations products and data services. We design industrial gateways, expansion modules and an AIoT platform in-house. There are two lines: standard product complement, adding connectivity to equipment and projects; and packaged solutions that take a scene from hardware through operations. Both share one device-edge-cloud foundation. The legal entity is Guangzhou Tongwei Smart Technology Co., Ltd.
How is Tongwei different from a sensor vendor or a dashboard vendor?
The difference is that we deliver a closed loop, not a single item. A sensor vendor solves measurement, a dashboard vendor solves visibility. Tongwei covers the whole chain: onboarding, alarms, dispatch, handling and review — including how protocols are integrated, who receives the alarm, who owns the fix and how the result is checked.
Which customer types does Tongwei serve?
Two primarily: equipment makers, cabinet builders, low-voltage contractors and OEM partners take product complement; integrators, maintenance providers, parks, public facilities, energy users and agri-aquaculture take packaged solutions. Energy, equipment rooms and restrooms are scenes inside packaged solutions, not a third business.
What is the relationship between "Zhuozhen Yunjing" and Tongwei?
Zhuozhen Yunjing is Tongwei's legacy brand and case asset in smart restrooms, now presented within the public facility operations line. The single external brand is Tongwei.
02
Technology and onboarding
Can our equipment be connected to the Tongwei platform?
It must be confirmed by assessment of the equipment list, comms interface, protocol document or point table, and site network and power conditions — we make no blanket claim to connect anything. Devices speaking Modbus, RS485/232, SNMP, MQTT or DI/DO are usually straightforward. Proprietary protocols require separate assessment, sometimes obtaining key states via DI/DO, analog inputs or serial traffic analysis.
Which communication protocols are supported?
Commonly Modbus RTU / Modbus TCP, RS485, RS232, SNMP, MQTT and DI/DO. Other protocols are handled through templated adaptation, scoped and quoted separately.
Does the system still work if the site network is unstable or down?
Yes. The gateway buffers data locally during an outage and backfills in time order once the link returns. Critical interlock rules run at the edge and do not depend on cloud connectivity. Buffer duration depends on sampling rate and point count, estimated against the real point table.
How many devices or points can one gateway handle?
There is no single figure. Capacity depends on protocol type, point count, sampling rate, bus length and device response time. Provide the point table and we will advise on per-unit capacity and distributed deployment.
Will onboarding interfere with existing control logic?
By default we read only and do not touch existing control logic. Where remote control is genuinely required, we first define safety boundaries, permission tiers, operation auditing and the responsibility split with the original control system. Tongwei does not replace field protection, fire interlocks or production control systems.
Is on-premise deployment supported?
SaaS, on-premise and private deployment are all supported. The choice follows data-security requirements, in-house operations capability and budget. Data ownership, access rights, tenant isolation and anonymisation are set out in contract.
Can we take only the data API and skip your UI?
Yes. Open APIs let data flow into your existing system, data hub or own front end. OEM branding is also available. Integrations and OEM pages are scoped and quoted separately.
03
AI capability and limits
How mature is Tongwei's AI capability today?
Verifiable today is L1 to L3 — visual monitoring, proactive alarms and closed-loop operations — plus selected verifiable L4 modules: trend anomaly detection, health scoring, report generation and knowledge Q&A. Predictive maintenance and automated AI diagnosis require sufficient history, fault labels and validated model performance before we promote them, and we do not present them as off-the-shelf features.
Is anomaly detection model-based or rule-based?
Both, but today mainly rules and statistics: thresholds, state checks, loss-of-contact detection, period-on-period comparison, trend deviation and cross-source correlation. These are more reliable with limited data and far easier to explain and verify with the customer. Model-based methods are applied where sufficient history already exists.
Will our data be used to train models?
Not for cross-customer model training by default. Data ownership, access rights, tenant isolation, anonymisation requirements and any model-authorisation scope are all set out in contract.
04
Delivery and commercial
What is the full engagement process?
The path is: diagnosis, selection / sample, single-site pilot, volume copy. Product complement emphasises sample integration then volume supply; packaged solutions emphasise a single-site pilot then multi-site copy. We favour a verifiable pilot over building a comprehensive platform up front.
Are samples and PoCs chargeable?
Chargeable in principle, with partial credit possible against later volume orders. Charging ensures both sides commit to a defined scope and acceptance target, rather than an open-ended free trial.
What does a quotation cover?
Standard hardware, platform licensing, deployment, cloud resources, SMS and data traffic, and ongoing operations are costed separately. Protocol adaptation, custom development (NRE), OEM pages, third-party integrations and advanced reporting are quoted after scope confirmation.
Can you commit to energy savings or fault-reduction percentages?
No. Such figures depend on the site baseline, equipment condition, operating habits and investment; a percentage detached from a specific project is meaningless. Our approach is to first make the consumption structure, anomalies and handling process visible, then verify results with comparable before/after data.
What is the typical timeline?
It depends on protocol complexity, point-table size, site count, field conditions and deployment mode, and is committed only after assessment. A single-site pilot is normally far quicker than a multi-site rollout.