Imagine a fictional “Acme manufacturing corporation”. Acme is a large, global company with interests in ship construction, chemicals, consumer goods, including plastics and food. This requires them to have a range of manufacturing approaches across their business, including discrete, continuous, and configure-to-order production.
Like a lot of large companies, Acme has a varied and complicated IT landscape with different business units using different production tools and software instances. This complicates getting insights into how different units are operating and has prevented digitalisation initiatives from delivering their full and expected benefits. It also limits their ability to leverage any benefits of scale.
Throughout our blog series, we’ll use this fictional corporation to demonstrate how our solutions can benefit a range of companies in different industries.
ISA-95 hierarchy
As our Industrial pyramid heavily builds on the ISA-95 functional model, let’s go through what that model is and the hierarchy it describes.
Most Factory Engineers understand what ISA-95 means, but it isn’t universal across enterprises. So, what is it and why should anyone from Acme who doesn’t work in the factory care?
ISA is the International Society of Automation and its 95 standard concerns developing an automated interface between enterprise and control systems.
What does this mean in business terms? Acme have adhered to traditional approaches where there has been a strict demarcation between domains. Corporate IT, which owns enterprise resource planning, product lifecycle management, and customer relationship management, is separate from factory IT. This area in Acme is responsible for systems such as SCADA, B2MML, machine-specific control systems and so on.
The ISA-95 functional model can be seen as a hierarchy that would allow Acme to integrate these systems to reduce the manual activities required to execute a process. This part that tends to receive less attention from businesses: factory IT contains those systems, such as MES (manufacturing execution systems), that collect data which allows an organisation to gain insight into their processes so they can make data-driven decisions.
Physical interaction layer
Having looked at the broader picture, we now move closer to the ground toward levels 0 to 2, where the physical production happens. These levels encompass the actual machines, sensors, and control systems that carry out and monitor the tangible manufacturing processes on the shop floor.
In day-to-day factory operations, it is responsible for executing physical actions, capturing real-time data from the environment, and responding to commands from higher-level control systems. It serves as the foundation for all automation and digital integration in manufacturing, acting as the essential interface between the physical world and the digital systems that monitor, control, and optimise production. Without the data and activity generated at this level, higher layers of the industrial hierarchy—such as control, execution, and enterprise planning—would lack the input necessary to function effectively.
While Level 0-2 focuses on physical processes and equipment, it’s essential to recognise that humans remain a critical part of the manufacturing environment. Operators, technicians, and line workers are often the ones who steer machines, perform manual tasks, monitor equipment, and respond to unexpected situations. Their experience and judgment contribute to maintaining quality, safety, and efficiency. When designing and integrating physical systems, it’s crucial to consider how people interact with machines through interfaces, physical access, ergonomics, and information flow. Ignoring the human role at this level can lead to inefficiencies, safety risks, and resistance to adoption. Therefore, systems should be built not just to automate, but to support and enhance human capabilities, ensuring seamless collaboration between people and technology on the factory floor.
For example, in many factories, operators are provided with screens intended to support their tasks, but the information displayed is often designed with machine-level diagnostics or engineering data in mind (such as raw sensor values, internal error codes, or system logs). While this data may be critical for automated systems or maintenance technicians, it can overwhelm or confuse frontline personnel who don’t need such granular details to perform their roles effectively. This misalignment can lead to frustration, reduced efficiency, and even operational errors, as workers may have to sift through irrelevant information to find what truly matters. Designing interfaces that prioritise clarity, context, and the actual needs of the human users is essential to ensure that technology enhances, rather than hinders, their work.
Many of the systems and interfaces at Level 0 are built with a very narrow focus—typically designed for a specific machine or a single part of the production line. While these interfaces may serve their immediate purpose well, they often lack visibility into the broader manufacturing context. This siloed approach means that operators and other personnel are limited to understanding only the local performance or issues of one machine, without insights into how their work fits into the full end-to-end process. As a result, coordination between workstations, proactive problem-solving, and overall process optimisation become more difficult. If these systems were designed with a holistic view of the entire manufacturing workflow—integrating data across machines, departments, and stages of production—they could provide much more valuable insights. This would support better decision-making and foster a shared understanding of the production goals, bottlenecks, and opportunities for improvement across the entire factory floor.
Building on this foundation, it’s also important to consider the role of emerging technologies such as audio analysis and add-on instrumentation. These solutions extend the sensing capabilities of machines beyond what is traditionally built-in, enabling more advanced forms of diagnostics and condition monitoring. It’s not just about measuring values that are critical to the machine’s immediate function; it’s also about analysing the meaning of those values in the broader context of production quality, process optimisation and maintenance. For instance, using audio fingerprinting, it’s possible to detect subtle anomalies and changes in engine performance that would be missed by conventional sensors alone, as shown in this exploration of internal combustion engine analysis.
Integrating this kind of add-on sensor data with operational control systems and business data opens new possibilities for predictive maintenance and intelligent automation. For example, in our work with a transmission system operator, sensor data from remote assets was combined with cloud-based IoT analytics to proactively manage maintenance needs, showcasing how new instrumentation can significantly enhance system-wide awareness and efficiency. Add-on sensors are often needed to reduce time to market for the diagnosis systems by postponing the need for deeper integration towards the operational systems of the factory.
Equally important is the question of how people interpret and act on this data. If the insights generated by add-on sensors or audio analysis aren’t accessible or understandable to the operators and maintenance teams, their value is lost. Human-centered design and visualisation play a key role here. In a project involving a modern sander machine, we helped enable the machine to “communicate” in ways that were meaningful to its users. Bridging the gap between complex technical diagnostics and day-to-day operational needs. It helps if the data that is being analysed is understandable for human beings in a way that, e.g. audio is.
By combining enriched sensor data with intelligent analytics and intuitive interfaces, predictive maintenance becomes more advanced and actionable. This approach shows how machine learning and advanced signal processing can surface actionable insights that help both machines and people work smarter. One of the simplest, however often overlooked, ways of enriching the data is to collect the situation information from the plant. That may mean, e.g. populating a simple characterisation of the situation along with measured data. That becomes necessary in further diagnostics and analyses.
In essence, enhancing Level 0-2 systems with add-on sensing and data integration isn’t just a technological upgrade; it’s a step toward more responsive, transparent, and human-aware manufacturing systems. This evolution makes factories more efficient, adaptable and resilient by combining the best of machine data and human judgment.
In summary, the most successful implementations consider the entire manufacturing process and the people involved in it. By designing systems that align with both technical requirements and the practical needs of operators, factories can achieve smoother workflows, better communication, and more responsive operations. This ultimately leads to more efficient and well-functioning production environments. Importantly, even the most intelligent systems must account for the human layer, because unlike machines, people bring ambitions, constraints, and lived realities that fundamentally shape how work gets done.
What does the future of physical production systems look like?
Looking ahead, the future of Level 0 in factory settings is marked by a clear shift toward greater connectivity, autonomy, and flexibility. At the core of this transformation is a move away from isolated machine control toward integrated machine-to-machine communication. In tomorrow’s factory, a machine won’t just perform its own task. It will be aware of its place in the process chain. For instance, if an upstream workstation encounters a problem or slows down, the following machines can automatically adjust their behaviour, avoiding jams, congestion, waste, or idle time. This kind of real-time responsiveness will drive smoother workflows and reduce the need for manual intervention.
Another significant trend is end-to-end automation. From raw materials to finished goods, systems will increasingly operate with minimal human coordination, enabled by faster data transfer, advanced control algorithms, and context-aware automation. The ability to reduce the amount of data generated and transferred will also be important. Data processing and transfer need to be appropriate. Key to this shift is the addition of new types of sensors and the rise of virtual sensors that infer machine states or process quality from indirect measurements. With advanced process control layered on top of these sensing capabilities, factories will be able to react to variations in real time. Whether they come from machine wear, product changes, or external disruptions. In this mesh of devices, the data that exists in any processing unit must be a bit smarter than the data that came in. That is the only way the amount and quality of data will be in any reasonable check.
Flexible manufacturing will also become central. In a future where customer demand is constantly changing, production processes must adapt quickly. Workstations mounted on mobile platforms, reconfigurable robotic systems, and software-defined control logic will allow factories to reshape their operations on the go. No longer constrained by fixed layouts, these systems can support a wide range of product variations with minimal downtime. This flexibility will also be supported by innovations like virtual PLCs, which can be configured and deployed remotely via the cloud, speeding up setup and changeover processes and reducing reliance on local hardware.
As AI systems increasingly become part of industrial processes, they must also learn to interpret the physical world in meaningful ways. To do this, we need to bridge the gap between machine learning models and the realities of production environments. One emerging requirement is the ability to embed physical context into the design of AI: to tell ML/AI what “physical” means or looks like. There are growing indications that the competencies required for building intelligent physical systems are shifting. Traditional disciplines like signal processing, physics, and control engineering are now gaining renewed relevance. When combined with modern analytical methods like machine learning, these fields provide the crucial context that AI systems need to interpret and operate effectively in the physical world. Their embedded, often tacit knowledge can offer ML models a more grounded sense of concepts. For example, even basic concepts like “upward” or “downward” are not self-evident to AI. But when systems are trained to consider real-world parameters, such as mass, acceleration, flow, inertia, and direction, they gain a better chance of performing effectively in complex, real environments. Classical science, in this sense, becomes a translator between physical reality and digital intelligence.
As these technologies evolve, user experience will become a differentiating factor. The best-designed systems will automate more and empower the people who remain in the loop. Operators will receive richer, more contextual data that supports faster decisions and more autonomous action. Interfaces will shift from technical readouts to intuitive tools that support real-time situational awareness, making it easier for users to understand what’s happening and take meaningful action. This improved experience will elevate the role of the operator from button-pusher to process leader.
Finally, the broader smart ecosystem surrounding Level 0—from facility infrastructure to energy systems to external service platforms will increasingly act as an integrated part of the production system. Sensor data won’t just stay within the machine; it will flow into enterprise systems, maintenance platforms, and digital twins, creating a smart environment where everything is aligned to support production goals. The result is a factory that runs faster or cheaper, that operates with greater intelligence, agility and resilience, and is built around the needs of both the process and the people running it.
Read more about Solita Industrial. And stay tuned for part 2 of our industrial pyramid blog series.
