In most companies, digitization has successfully arrived. The processes are primarily digital and products have at least one digital component. Processes, applications and interactions generate digital data, or at least have the potential to do so. But how can the resulting data be used sustainably and ethically?
This data is often used to analyse and optimise business processes. How would it be, however, if data evaluation could also be used beyond that to enable new offers and products in the field of artificial intelligence (AI)?
The value of the data for this purpose often remains hidden when viewed without context. For example, a comprehensive view of a company and its existing competencies and development potential is required to find suitable approaches for AI-driven business models.
It is sensible to welcome this technological development. Due to the existing digital infrastructure and database, innovative business models and value chains based on AI will emerge much faster than digitalisation required for their introduction.
But how should an innovative business model based on artificial intelligence be designed in order to be successful in the market?
Artificial Intelligence from the Developer's Perspective
Andrew Moore, Dean of the renowned School of Computer Science at Carnegie Mellon University, summarises the definition of AI in one sentence as follows: "artificial intelligence refers to the [...] construction of methods that enable computers to perform tasks that were previously thought to require human intelligence."  This definition remains deliberately vague and does not specify the methods to be used.
Most implementations of artificial intelligence are currently based on so-called machine learning. This is an umbrella term for a class of computer algorithms that can automatically improve their functionality and performance through experiential learning  This includes not only the famous neural networks, which provide the most spectacular advances in the field of AI (and whose application is also called deep learning because of their internal, multi-layered structure), but also long-established methods of mathematical statistics such as decision trees and forests, which are now finding new applications.
Artificial Intelligence from the User's Perspective
Nevertheless, real artificial intelligence needs more than just a self-learning algorithm. The AI must be able to interact with its user in order to react flexibly and dynamically to his needs. However, it must also be able to act autonomously in order to fulfil a pre-defined purpose.
A well thought-out and human-centred design of the AI application is required in order to gain the necessary trust of the user in their abilities. Familiar software development tasks, such as connection to external services and embedding in existing or new digital ecosystems, are also part of this in most cases.
Based on these considerations, it becomes clear that data evaluation does not necessarily require machine learning, but that even functionality provided by machine learning is not automatically an artificial intelligence.
Automation through Machine Learning
Common approaches to the use of AIs that are often discussed concern the automation of monotonous and repetitive processes. Many of these were previously reserved for humans due to a certain complexity in their implementation, but can now be implemented by computers or machines using simple machine learning algorithms.
The major incentives for development are the resulting increase in efficiency and cost savings in the automated handling of these processes. The optimization of business goals through better data evaluation and forecasts of relevant KPIs are also frequently mentioned.
Unfortunately, these applications - and they will certainly be used over time - generate fear of AI solutions due to the threat of job loss and social acceptance is reduced. Above all, however, even only slightly innovative applications of a fascinating and promising technology offer enormous potential for completely new business models.
Innovation through Artificial Intelligence
Actual innovation through artificial intelligence can only come about through a completely new business model. The traditional image of vertical integration between production and retail is transformed into a completely digital, AI-based value chain:
Since conventional business models are often attacked or become redundant due to corresponding offers, it makes sense in this context to speak not only of vertical offer integration, but even of vertical disruption.
On closer inspection, the core concept of disruption through vertical integration can be found in almost all innovative and successful AI-based business models. In this context, Bradford Cross coined the following basic ideas: 
1. Generating and collecting proprietary data that is difficult for competitors to reproduce. For this purpose, for example, an attractive product for consumers can be created or an existing, highly specialised product such as an industrial machine can be used.
2. The data flow into a machine learning algorithm underlying artificial intelligence. Having control over the data collection means the exact data can be selected that guarantee optimal functionality. If necessary, the quality of the data can be readjusted at any time. For the same reasons, the interface to the user must also be part of the offer: in order to be able to deliver the greatest possible benefit - but also in order to be able to continuously adapt the offer to changing user needs.
3. The functionality achieved through machine learning and the underlying business model should be based on great expertise in the chosen industry. Only then can the result of AI functionality become a novel offering that not only meets the needs of users, but also has sufficient confidence in its capabilities to achieve high market acceptance.
The Market is full of Opportunities
Why is vertical integration and disruption so important for AI offers? An appropriate business model that encompasses the entire digital AI value chain puts the provider in a position to offer the user unique added value and generate the complete return on the solution itself. At the same time, however, high barriers are also being set for new competitors.
A company that wants to offer an AI-based product must completely master this chain and base its business model on AI-based core value creation. Only then can user needs be sustainably met which would not have been conceivable or feasible without AI technology.
The individual components of the value chain are not a suitable basis for a particular position in the market. Rather, the completion of routine tasks very quickly becomes a commodity. Furthermore, in software-based markets in particular, today's innovation is tomorrow's open source more than ever.
Only through complete vertical integration, from the collection of proprietary data to the use of current machine learning technology to the use of domain-specific expertise, can artificial intelligence achieve sustainable added value for the user and thus a disruptive offer can be realised.
Let's talk about the opportunities your expertise and data collection potential can offer for your AI-based business model of tomorrow.
- Forbes, Carnegie Mellon Dean Of Computer Science On The Future Of AI (30. Okt. 2017)
- Tom Mitchell, Machine Learning, McGraw Hill (1991)
- Bradford Cross, CEO of CEAi – Vertical AI Studio