Why Established Companies Can Develop Successful AI Products but Still Fail at Business Model Innovation
​

While product innovation addresses the question of what customer value is created, business model innovation focuses on how this value can be commercialized, scaled, and captured.

 

Although the distinction appears straightforward in theory, it is often neglected when companies pursue innovation initiatives. In the context of AI, however, it is frequently underestimated. Many companies invest substantial resources in developing new products without questioning whether their existing business models are capable of successfully commercializing the newly created customer value. This is where the core argument of this article begins.

Why Established Companies Possess Strong Foundations for AI Products

Public discussions about artificial intelligence often focus on AI models, computing power, and technological capabilities. This can easily create the impression that developing successful AI products is primarily a technological challenge. Such a perspective, however, overlooks a number of critical factors.
 

Successful AI products require far more than powerful AI models. They depend on a deep understanding of customer problems, domain expertise, expert knowledge, and access to data that describes a specific use case and is not easily available to competitors. This is exactly where many established companies have a significant advantage.
 

Industrial manufacturers often own decades of operational and service data from their installed equipment base. Healthcare companies possess extensive diagnostic knowledge and application expertise. Industrial businesses frequently understand their customers’ processes, challenges, and decision-making logic better than ex-ternal technology providers. These resources are often considerably more difficult to replicate than the underlying AI technology itself.
 

At the same time, technological barriers to entry have decreased significantly over the past few years. Powerful AI foundation models are now available through both commercial and open-source offerings, allowing companies in many cases to avoid developing or training their own foundational models. The real value often emerges through adapting these models to specific use cases and combining them with proprietary data and existing expertise.
 

As a result, the primary constraint is increasingly no longer access to technology. Instead, it is the ability to translate existing knowledge and available data into marketable products. This trend is further supported by the ability to keep sensitive data within corporate boundaries. Modern AI solutions can be deployed within private cloud environments or on company-owned infrastructure, reducing concerns regarding data privacy and intellectual property protection.
 

For many companies, the key challenge is therefore not whether AI can be applied, but rather how existing assets can be identified and transformed into new products. This is precisely why many established businesses are in a stronger position than they often realize themselves.
 

From Data and Knowledge to AI Products

The practical implications of this development can already be observed across a variety of industries.
 

John Deere provides a compelling example. Over several decades, the company has accumulated extensive data related to agricultural processes, machinery, soil conditions, and farming practices. This data was not originally collected with the intention of creating AI products. It emerged naturally as part of the company’s core business. Only through the combination of this proprietary data with modern AI technologies did solutions such as See & Spray become possible.
 

From the customer's perspective, AI itself is not the primary source of value. Farmers are not purchasing image-recognition technology. They are purchasing the ability to apply agricultural inputs more precisely, reduce costs, and maintain crop yields.
 

A similar pattern can be observed at Siemens. The company possesses decades of expertise in industrial processes, engineering, and automation. This knowledge serves as the foundation for AI-based products such as the Industrial Copilot. Once again, the company is not primarily selling technology. The customer benefit lies in accelerating complex tasks, increasing productivity, and improving the utilization of existing resources.
 

Roche demonstrates the same underlying logic within an entirely different industry. The company combines medical expertise, diagnostic data, and clinical experience with modern AI technologies to develop new diagnostic solutions. The primary value does not reside in the AI itself, but in the ability to improve medical decision-making and enhance diagnostic processes.
 

Despite operating in different industries, these examples share a common pattern. The starting point was neither a new AI model nor a breakthrough technology. The starting point was existing data, knowledge, and customer access. AI was subsequently used to transform these proprietary assets into new products and new forms of customer value.
 

The examples also highlight a second important observation. None of these companies limited themselves to developing an AI product alone. New capabilities were built, additional partners were integrated, and existing value creation structures were expanded. Even at this stage, it becomes apparent that successful AI products often extend beyond pure product innovation and require adjustments to other elements of the business model.
 

This highlights one of the most important insights for established companies. The development of successful AI products rarely begins with the question of which AI technology should be used. A far more relevant question is which knowledge, data, and customer problems already exist and how they can be transformed into a differentiated product.
 

Why AI Products Create a Different Type of Customer Value

Historically, many industrial companies have differentiated themselves through the characteristics of their physical products. Performance, quality, precision, reliability, and durability have been at the core of value creation and have formed the basis of competitive advantage for decades.
 

These factors will remain important in the future. AI, however, adds an additional layer of differentiation. Customer value is increasingly created through information, predictions, and decisions. Companies are therefore no longer selling only a physical product. They are increasingly selling knowledge, recommendations, optimizations, and outcomes.
 

A machinery manufacturer, for example, no longer sells only a machine. It can additionally offer the ability to predict failures, optimize maintenance activities, or increase equipment availability. A diagnostics provider no longer sells only a diagnostic system. It can also improve the quality of diagnostic decisions. An agricultural company no longer sells only machinery or agricultural inputs. It can additionally provide recommendations for the optimal management of farmland.
 

As a result, customer value changes not only quantitatively but also qualitatively. Customers are not simply receiving more of the same value. They are receiving a new form of value that was previously unavailable.

 

This development is particularly important for established companies. AI often extends existing products with new digital value components that are based on data, knowledge, and predictive capabilities. Differentiation therefore shifts, at least in part, from the physical characteristics of a product to the intelligent utilization of that product.
 

This is where product innovation and business model innovation begin to intersect. When customer value changes, the conditions required to successfully commercialize, deliver, and monetize that value often change as well.
 

The Innovation Trap of Many Established Companies

Many companies are currently developing new AI products. Far fewer are questioning whether their existing business models are capable of successfully commercializing and scaling the customer value these products create. This represents one of the most significant challenges of the coming years.
 

AI products frequently create a different type of customer value than a company’s traditional offerings. Nevertheless, they are often marketed through the same sales structures, incentive systems, and organizational mechanisms that were originally designed for physical products or traditional services. This creates a disconnect between the customer value being created and the strategic conditions required to successfully bring that value to market.
 

The consequence is not necessarily product failure. More commonly, the economic potential falls short of expectations. A product may be technologically successful while the company lacks the necessary conditions to translate the created customer value into sustainable competitive advantage and economic success.
 

Established companies are particularly exposed to this risk. Their business models have often been optimized over many years for existing products, customer relationships, and market dynamics. AI products, however, frequently follow different economic logics and affect multiple business model elements simultaneously. The real challenge therefore lies not only in developing the product itself, but in adapting the strategic and organizational conditions required for its long-term success.
 

This challenge is often underestimated because discussions around AI remain heavily focused on technical feasibility. Yet once a successful product has been developed, the considerably more difficult task begins: establishing that product sustainably in the market. This is where product innovation and business model innovation intersect.
 

Why Successful AI Products Often Require Business Model Changes

The most visible impact often emerges within the sales organization. A traditional machinery sales force sells machines, spare parts, and service contracts. Customer discussions revolve around technical specifications, performance characteristics, availability, lifespan, and capital expenditure.
 

AI products introduce entirely different conversations. Customers become interested in the origin of the data, the quality of the underlying models, integration into existing system landscapes, data privacy considerations, and the expected economic outcomes. Consequently, not only the product changes, but also the nature of the sales process itself.
 

Many sales organizations are only partially prepared for these changes. This is not due to a lack of commitment or capability. Rather, new products often require new skills. Sales professionals who have successfully sold physical products for many years do not automatically possess the expertise required to explain complex data-driven solutions or quantify their economic value.
 

A further challenge, often underestimated in practice, relates to incentives. Sales organizations typically operate within established compensation structures. These structures have been optimized around existing products and reflect the economic priorities of the business. New AI products therefore compete for attention with established offerings, even though their market potential, margin profile, and probability of success may still be uncertain during the early stages.
 

From a corporate perspective, these products may be strategically important. For individual sales representatives, however, it may appear economically rational to continue prioritizing the products with which they have already been successful. The challenge therefore lies not in the technology itself, but in the design of the business model and the associated incentive structures.
 

Similar questions arise regarding resources and capabilities. Many AI products require capabilities that have historically been limited within industrial companies. Examples include product management, customer success management, data management, software operations, and the continuous enhancement of digital products.
 

Whereas physical products are often updated only through larger development cycles after delivery, AI products evolve continuously. Customers expect regular improvements, new functionality, and ongoing optimization of results. Product development therefore does not end with the sale. It becomes part of ongoing operations.
 

This also changes the underlying value chain. Activities that previously played only a minor role become increasingly important. At the same time, new requirements emerge regarding organizational structures, processes, and responsibilities.
 

The value network often changes as well. Traditional industrial companies typically collaborate with established suppliers, distributors, and service partners. AI products frequently require additional collaboration with cloud providers, software integrators, data partners, and technology platforms. As a result, the existing network expands to include new actors that become relevant for product development, delivery, and scaling. In many cases, the target customers and decision-makers associated with the AI product also differ from those of the traditional offering.
 

The development of successful AI products can therefore require adjustments across nearly all elements of a business model. These include:
​

• Resources and capabilities

• Value chain

• Cost and revenue structure

• Addressed industries, markets, and market segments

• Value network

• Business field strategy

 

The more the newly created customer value differs from a company’s existing offerings, the greater the business model adaptation required to successfully commercialize it.
 

ARVIEN Product and Business Model Innovation Matrix

The observations discussed so far can be summarized in a simple matrix. It is based on two dimensions of innovation that are critical to the long-term success of AI products.
 

The first dimension describes the degree of product innovation. It addresses the extent to which a company develops new products or new forms of customer value.
 

The second dimension describes the degree of business model innovation. It addresses the extent to which the strategic conditions required to successfully commercialize and economically capture that customer value are adapted.
 

The combination of both dimensions results in four fundamental situations.