How development changes when products become cyber-physical systems

For many organisations, this creates a fundamental challenge. Teams continue to operate in silos, software remains a supporting function rather than a strategic driver, and integration comes too late in the process. The result is rising complexity, slower delivery, and underwhelming user experiences.

In the first article in our series on cyber-physical systems, we explored what this transition means for industrial companies: how the shift from physical products to integrated digital ecosystems is redefining where value is created, how competitive advantage is built, and how products are evolving into something fundamentally new.  

But as products evolve into interconnected systems, the way they are developed must evolve accordingly. In this second article, we examine how cyber-physical systems are reshaping product development and why delivering them requires a different approach to engineering. 

In traditional industrial products, value is created primarily by the performance of their physical components: the precision of a bearing, the torque of a motor, or the durability of a chassis. Engineering excellence comes from how well those components are designed and manufactured.

Cyber-physical systems change that logic entirely. Here, competitive advantage no longer resides in individual components, but in how software, connectivity, data, and user experience interact. 

No single component delivers value by itself. The system itself becomes the product.

This changes the basis of competition. Manufacturers can still build outstanding hardware and lose market share to competitors that deliver better system integration, faster software evolution, or superior digital experiences.  

For users, the difference between a cohesive solution and a fragmented product is immediately obvious. Well-designed systems feel seamless, with physical and digital touchpoints that reinforce each other, intuitive workflows, and updates that improve the solution over time. Fragmented products, by contrast, create friction through disconnected and inconsistent digital interfaces.

When value emerges from the system as a whole, isolated component engineering is no longer enough. As a result, industrial companies must rethink not only the products they build, but also how they build them. 

A fundamental shift in cyber-physical systems is that product development no longer ends at launch.  

Conventional hardware products followed a linear lifecycle. A product was developed, released on a specific date, incrementally improved through service updates, and eventually, teams pivoted and replaced it with the next generation. Because functionality remained relatively stable after launch, development could largely operate through sequential handovers between disciplines and late-stage integration.

Cyber-physical systems change those assumptions. They are never truly complete, as software upgrades extend utility, operational data surfaces new behaviours, and AI models introduce new autonomous functions.

As products continue to evolve after deployment, traditional development models are increasingly difficult to sustain. Sequential handovers between teams grow more fragile as decisions in one area ripple across the entire system. With rising complexity, integration begins to expose interface gaps and hidden dependencies that are difficult to detect early. The later these issues surface, the more costly and disruptive they are to resolve

Development shifts from delivering finished products to managing continuously changing systems. As a result, several development practices become essential: 

Traditional product development begins with components and integrates them into a finished product. Cyber-physical systems development starts with the intended outcome and designs the system to deliver it. When value is created through system behaviour, architecture becomes the central discipline, rather than a background engineering concern.  

In practice, architecture becomes one of the primary determinants of whether products can evolve successfully or descend into growing complexity. This requires clear links between user needs, system requirements, and implementation decisions so that functionality, trade-offs, and costs remain aligned throughout the product lifecycle.

To support this evolution, architectures must provide:

• Clear traceability from business goals to system requirements  
• Interfaces that allow teams and subsystems to evolve independently
• Modular functions that combine into customer-facing features  
• Low coupling between subsystems and strong cohesion within them

These principles allow cross-functional teams to work independently while maintaining overall system coherence. Without them, integration slows, dependencies accumulate, and product changes become harder to deliver reliably. Teams that bypass proper system architecture because they “need to move fast” often face the consequences later through escalating costs, integration challenges, and reduced responsiveness to changing consumer needs.

Architecture is therefore the foundation for sustainable adaptation and integration over time. As systems become increasingly software-defined, technologies such as AI play a growing role in shaping product capabilities, behaviour, and experience. 

AI amplifies many of the changes already transforming cyber-physical systems. Importantly, it creates value through coordinated product behaviour rather than as an isolated feature

For example, an anomaly detection model is only valuable if it can access the right sensor data, integrate with maintenance workflow, and trigger the appropriate physical response. It depends on strong system architecture, seamless integration between components, and a clear understanding of how the overall system solves meaningful customer problems. Organisations that treat AI as an isolated software feature often discover that these foundations are missing.

Designing AI-enabled products starts with intended outcomes rather than standalone features. Because future requirements remain inherently uncertain, organisations must anticipate shifts in users, markets, and operating environments.

This becomes even more important as AI capabilities evolve rapidly. Organisations can no longer assume that software functionality will remain stable across long hardware lifecycles. Products increasingly need architectures capable of absorbing future AI capabilities that do not yet fully exist.

AI is also reshaping how intelligent systems are developed. Used effectively, it can support systems engineering through automated consistency checks, architecture reviews, dependency analysis, and integration risk assessment. When system relationships are clearly defined, AI can propose mitigations and perform plausibility checks across requirements and implementation layers.

But despite rapid technological progress, many companies still approach AI and cyber-physical systems with outdated assumptions on product development. Over time, that disconnect becomes a growing competitive risk. 

For companies and product organisations building complex systems, the question is no longer whether to adapt, but how quickly to do so. Five priorities are critical to making that transition successfully.

Industrial companies spent decades optimising development around predictability, stability, and component excellence. Increasingly, those assumptions are no longer sufficient. In cyber-physical systems, complexity does not remain static after launch. Products continue adapting, dependencies continue growing, and integration becomes a permanent engineering challenge rather than a final development phase. 

The real transformation is not only technological, but organisational. Building cyber-physical systems ultimately requires development models designed for iterative adaptation rather than stable, isolated products.