The Label Problem on the Shop Floor
Digital twin is used for almost everything. A CAD file gets called one. A full physics simulation gets called one. A data lake with a dashboard gets called one. An AI forecast model gets called one. When a plant team sits down to evaluate tools, that looseness creates real friction. Scope, timeline, and day-to-day usefulness all shift depending on which meaning the vendor is using.
One practical way to cut through the noise is the market segmentation laid out in the Twinzo presentation. It separates five categories that sit on a clear progression: static, simulation, data-platform, operational, and predictive. Each answers a different question. Mixing them up is how projects end up with the wrong system for the actual problem.
Static Digital Twins
Static digital twins are the tools that produce the geometric model itself. CAD packages, architectural modeling software, gaming engines used for visualization, LiDAR scans, photogrammetry, and related methods all live here. The output is a three-dimensional representation of the facility, machines, or layout. It does not move with live data. It does not answer operational questions on its own. It is simply the geometric base that later layers can reference.
Most plants already have pieces of this layer. Layout drawings, machine models, and occasional scan data exist in different systems. The static twin is useful for design reviews, training visuals, and as the spatial container for other data. It is not, by itself, an operational system.
Simulation Digital Twins
Simulation digital twins take a static model and place it inside a physics or discrete-event engine. The goal is to test what-if scenarios before any physical change happens: adding a warehouse, buying a new robot, changing headcount, or rearranging a production line. These tools explore possible futures under controlled assumptions.
They are strong for capital planning and layout decisions. They are not designed to reflect the live state of the floor. Once the simulation run is finished, the model sits still until the next scenario is loaded. Treating a simulation twin as a live operational view creates false expectations about real-time visibility and corrective action.
Data Platforms
Data platforms sit in the middle as the organizers of heterogeneous streams. MES systems, data-pipelining tools, and similar platforms attempt to create digital threads by collecting and structuring information from many sources. They focus on data organization and manufacturing intelligence rather than spatial context or real-time visual operations.
A good data platform makes production metrics, downtime events, and quality data available in a structured way. It does not automatically place those numbers on a live 3D map of the facility or show where a forklift is right now. That spatial and operational layer belongs to the next category.
Operational Digital Twins
Operational digital twins connect real-time data to a spatial model so teams can see what is happening right now and understand why. Position data from RTLS, production metrics from MES or ERP, IoT sensor values, and other live feeds are overlaid on the 3D representation. The purpose is corrective action on the current state, not prediction of future states or pure simulation of alternatives.
Twinzo sits in this operational category. It is built as a platform for live operational digital twins of industrial facilities. It ingests data from systems such as RTLS, MES, ERP, WMS, and SCADA and places that information in spatial context so operators and supervisors can act on what is actually occurring. According to the roadmap language in the Closed Loop DT materials, Twinzo is only directionally moving toward predictive capabilities; those features remain future-tense rather than shipped.
The distinction matters for daily use. An operational twin answers questions like “where is the material right now,” “why is that line waiting,” or “which forklift has been idle longest this shift.” Those are present-tense questions. Simulation and predictive tools answer different ones.
Predictive Digital Twins
Predictive digital twins apply advanced AI and machine-learning models to spatially oriented datasets. The aim is to estimate what may happen, surface correlations or causal relationships, and help prevent negative outcomes. The market is heading in this direction, and Twinzo’s own AI work is described as moving step by step toward it. No industrial solution is presented as having fully achieved the complete predictive layer yet.
This category sits furthest from today’s day-to-day floor decisions. It requires the spatially oriented dataset that an operational twin can help build, plus mature models that can turn that data into reliable forward-looking insight. Until those models are proven in the specific plant context, the practical value remains limited.
Keeping the five categories separate prevents the common mistake of buying a tool that solves yesterday’s problem or next year’s problem while the current shift still needs better visibility into what is happening right now. Static models give geometry. Simulation explores options. Data platforms organize streams. Operational twins show the live state. Predictive twins try to look ahead. Each has its place. The taxonomy simply makes the place visible before the purchase decision is made.