
Competitive advantages with knowledge-based AI
In the past, there has probably never been a battle for competitive advantage that has been as dynamic as the current race in the field of artificial intelligence (AI). New opportunities are arising for Europe with knowledge-specific (domain-specific) AI. These opportunities build on the traditional strengths of the „old continent“. In order to catch up, it seems necessary to take a closer look at the topic of knowledge and its long history of development.
As part of our series on AI as a tool for strategies, this new blog post follows on from my explanation of strategic learning loops. First of all, it deals with the combination of knowledge management and AI technologies in the context of the fifth development stage of connective strategic management.
Battle for leadership in AI
The five companies with the highest market capitalization worldwide (as of December 2024) are Apple, Nvidia, Microsoft, Amazon and Alphabet. Artificial intelligence is an important value driver. Apple is worth 3.7 trillion euros. All 40 DAX companies together are only worth 1.9 trillion euros.1
At the end of January 2025, the Chinese start-up Deepseek surprised the global public with a new AI language model that is said to be able to compete with the best models from Western tech giants, but requires less computing power and costs less. The news triggered a slide in US technology stocks. In the meantime, share price losses amounted to one trillion US dollars. The company, founded by Liang Wenfeng in 2023, relies on open source, i.e. the software is freely available to others. It is also said to have been trained without high-tech chips. This raises the question of whether the billions invested by US companies are really necessary. Deepseek’s good price-performance ratio is probably the result of a combination of different approaches, e.g. the composition of many small expert models, of which only the relevant ones are activated.2 For European AI providers with less capital strength, this development may represent an opportunity.
Competitive advantages with AI from Europe
When it comes to artificial intelligence, Europe faces the task of catching up and reducing its dependence on large tech companies. It is also important to secure critical infrastructures and protect the intellectual property of organizations based here. This is particularly important for the many hidden champions and their outstanding expertise in specialist areas. After the hype and some disillusionment with large language models, new opportunities are now emerging for an AI strategy that builds on the strengths of the European economy. Knowledge-specific (domain-specific) artificial intelligence plays an important role in this , providing competitive advantages for many small and medium-sized companies. The Heidelberg-based start-up AlephAlpha has developed a new approach to this.
The advantage for companies is that they can design and operate language models with their own knowledge. Today’s models are based on the transformer architecture and a tokenizer that recognizes language patterns. For this purpose, large volumes of text are analyzed and broken down into individual components (text segmentation). AlephAlpha’s T-Free approach and its AI model Pharia work differently. T-Free stands for tokenizer-free and continuously processes groups of three adjacent characters. This makes it easier to adapt to other languages and terminologies. Together with the semiconductor manufacturer AMD and the Finnish start-up SiloAI, which was acquired by AMD, AlephAlpha has found a way to train industry- and company-specific terms („languages“) with significantly improved performance using T-Free. The approach also helps to increase AI sovereignty(3).
A consortium of companies, universities and supercomputing centers is currently developing an AI for Europe. Peter Sarlin from SiloAI sees the new Open Europe LLM project as a „moonshot“. Participants from Germany include AlephAlpha and the Fraunhofer Institute for Intelligent Analysis and Information Systems (IAIS). Both the code and the research will be published as open source. The European Commission is to provide up to 54 million euros over the next three years. In international comparison, this sum is relatively small. However, a European AI that becomes a public good will significantly increase sovereignty(4).
According to experts, Europe has the opportunity to gain a competitive edge in artificial intelligence if it succeeds in combining the following four success factors:5
- Improved cooperation between politics, science, business and society
- a focus on knowledge-specific AI applications
- pooling resources to overcome disadvantages of scale and
- the creation of trustworthy AI as a differentiating feature.
Such a combination requires connective strategic AI management. While French President Emmanuel Macron wants to invest 150 billion euros in European AI start-ups, the topic of artificial intelligence is unfortunately barely mentioned in the German parliamentary election campaign.
In the following, I would like to outline how knowledge management can be successfully combined with AI technologies.
Combining knowledge management with AI technologies
The collaboration between humans and AI works in a similar way to pole vaulting. The pole is a tool that enhances the jumper’s abilities if they master the tool.
The use of AI changes people’s knowledge work in the following three dimensions:
- Time savings through automation of routine activities
- Expanding skills in processing both data-intensive and unstructured tasks and
- individualized learning for the further development of human skills.
The potential of AI as a tool lies in the interaction of these dimensions.
With the combination of knowledge management and AI technologies, a new way of achieving competitive advantages is now emerging. The starting point is the activation of companies‘ specific knowledge and skills. Added to this is the use of the potential of AI to expand competencies and thus to differentiate themselves from the competition. The third and decisive point is the systematic improvement of skills that combine knowledge and AI. This requires targeted training and further education.
For Jeanette zu Fürstenberg, Head of Europe at the US investment company General Catalyst, the opportunities for the European economy lie in combining the big data and knowledge of established companies with AI technologies(6).
In the following, I would like to explain the connection between knowledge management and artificial intelligence and discuss the implications for strategic management in this and the next blog posts.
Knowledge for innovative business models has a long history of development from ancient Greece to today’s knowledge society. In the 1990s, the conviction prevailed that the creation of new knowledge is an important source of competitive advantage. However, this hype surrounding knowledge management was followed by disillusionment. At the same time, US start-ups succeeded in linking knowledge-based value creation and value enhancement with digital business models.
The development of AI technologies has progressed from symbolic AI and neural networks to generative AI (GenAI). In 2024, four AI researchers were awarded Nobel Prizes. But the hype surrounding large language models is turning into disillusionment. Small language models have a number of advantages. They are cheaper and easier to adapt to specific applications. Here, too, the question arises as to how the dangers of AI can be contained.
The combination of these two topics leads to the realization that knowledge-specific AI is an important process and design element in strategies. A distinction can be made between the corporate strategy level and the functional strategy level. AI is a new tool for supporting strategy processes and the collaboration of strategy teams. In addition, AI enables the design of innovative products, services and business models. At the functional level, AI makes important contributions to increasing the productivity of connected business processes. In addition, AI-supported, agile performance management leads to better complexity management than traditional approaches.
Since knowledge management forms a basis for the use of AI technologies, I would first like to outline the development of the topic of knowledge from ancient Greece to the knowledge society.
From ancient Greece to the knowledge society
In ancient Greece in the 3rd century BC, the philosopher Plato and his student Aristotle discussed the question of whether deductive or empirical theories of knowledge lead to the acquisition of knowledge.
At the beginning of the modern era, Rene Descartes (1596-1650) propagated a separation between the subject of knowledge and the object of knowledge. This so-called Cartesian division was to occupy science for a long time to come.
The German philosopher Immanuel Kant (1724-1804) attempted a synthesis. Logical thinking and experience work together.
At the beginning of the twentieth century, American pragmatism, with representatives such as William James, was concerned with the relationship between knowledge and action.
In 1969, Peter Drucker coined the concept of a knowledge society characterized by knowledge work and knowledge workers(7.
The work of Chris Argyris and Donald Schön on single loop and double loop learning,8 which formed the basis for the concept of a learning organization, has been of great practical relevance since the late 1970s.
Surprisingly, the topic of knowledge did not play a decisive role in the resource-oriented view of strategic management that emerged in the early 1990s.
From today’s perspective, we define knowledge as a resource and the result of learning processes that people create in exchange with teams, organizations and artificial intelligence.
Creation of new knowledge as a source of competitive advantage
The concept of implicit or tacit knowledge, which the natural scientist and philosopher Michael Polanyi coined back in the 1950s, is important for the creation of new knowledge.9 In the case of tacit knowledge, someone knows how to do something, but their knowledge is implicit in their skills. It is difficult to document verbally or in writing in the form of explicit knowledge.
In the mid-1990s, Japanese scientists Nonaka and Takeuchi described how new knowledge as a source of competitive advantage arises from the following four forms of knowledge exchange:(10
- From implicit to implicit (socialization)
- from implicit to explicit (externalization)
- from explicit to explicit (combination) and
- from explicit to implicit (internalization).
These forms of knowledge exchange are crucial to the success of hidden champions. The combination of knowledge, skills and action has a long tradition there. The creation of new knowledge, the development of skills and their implementation in practical action often take place in learning processes in which – similar to sport – demonstration and imitation play an important role. These learning processes can be documented and scaled using videos, for example.
This provides new impulses for the application of knowledge-specific artificial intelligence
Hype and disillusionment in knowledge management
In the second half of the 1990s, knowledge management experienced a hype phase, which was followed by disillusionment. The hype was mainly triggered by the book The Knowledge Creating Company by Nonaka and Takeuchi, which deals with knowledge management in Japanese companies.
Ultimately, however, the importance of tacit knowledge has not really been understood „in the West“. The focus of companies and consultants has been on extracting and synthesizing existing explicit knowledge („if HP knew what HP knows…“). This proved to be difficult and costly and contributed to disillusionment in the 2000s.
A pragmatic approach that linked knowledge, skills and action did not play a major role in the publications of the time.
Knowledge-based value creation, value enhancement and AI-based business models
Knowledge-based value creation, value enhancement and the connection with digital business models are the subject of our book WissensWert (KnowledgeValue), published in 2001.11 Work on this began in the mid-1990s, inspired by the increasing importance of knowledge management. It followed on from the „reengineering wave“ and IT-based innovations in routine processes. Our initial hypothesis was that knowledge-based value creation and value enhancement with knowledge open up new opportunities for achieving competitive advantages.
At the same time, new digital business models have emerged with internet technologies, initially in online retail (electronic business). Following the collapse of the new economy, start-ups such as Amazon, Google and Facebook have achieved leading market positions and have become the most valuable companies in the world.
Europe has become heavily dependent on digital business models. Looking back, it is astonishing how little people here have noticed that business model innovations based on AI applications have emerged since the turn of the millennium.
In the early 2000s, then Princeton computer science professor Fei-Fei Li began building the largest database in AI research (Computervision, later ImageNet). One user was the online bookseller Amazon. Founded in 1994, the company is regarded as the inventor of AI-based personal product recommendations.12 Since 2003, Amazon has been using the item-to-item collaborative filtering method for this purpose.
Another AI user was Facebook with a social network that uses machine learning to bring people together („matching“) who have things in common. Machine learning models sort personalized advertising according to the highest probability of success, thus establishing innovative business models such as Google’s search engine and its RankBrain algorithm. Spotify’s music streaming business model, Netflix’s video streaming and the short video platform of the Chinese Bytedance subsidiary TikTok are also based on the AI-based principle of personal recommendations.
This means that many people have been in daily contact with AI applications since the turn of the millennium without realizing it. Europe is currently facing the challenge of making better use of the new opportunities offered by AI than in the past.
Conclusion
- In view of the extreme competitive dynamics in artificial intelligence, Europe must catch up and reduce its dependency
- One way to do this is to combine company-specific knowledge with innovative AI technologies such as the tokenizer-free approach
- The success of today’s tech giants since the turn of the millennium is based on the creative application of
- Knowledge-specific artificial intelligence could build on and continue the success story of the European hidden champions.
Literature
[1] Sommer U., USA dominates like never before. In: Handelsblatt, December 27/28/29, 2024, p.1, 6-8
[2] Gusbeth, S. et al, Sputnik moment. In: Handelsblatt, January 31, February 1-2, 2025, pp. 50-55
[3] Holzki, L., Up to 400 percent more efficient. In: Handelsblatt, January 22, 2025, p. 23
[4] Holzki, L., 54 million for a European AI. In: Handelsblatt, February 4, 2025, p. 18-19
[5] Bomke, L., Knees, L., Wo Europa Chnacen im KI-Rennen hat. In: Handelsblatt, February 10, 2025, p. 20-21
[6] zu Fürstenberg, J., „We need much more capital that also takes risks“ (Interview), In: Handelsblatt, January 31, February 1-2, 2025, pp. 32-33
[7] Drucker, P.F., The Age of Disconinuity – Guidelines to our Changing Society Butterworth-Heinemann 1969
[8] Argyris, L., Schön, D.A., Organizational Learning – A Theory of Action Perspective, Addison Wesley 1978
[9] Polanyi, M., Implicit Knowledge, Suhrkamp 1985
[10] Nonaka, I., Takeuchi, H., The Knowledge-Creating Company – How Japanese Companies Create the Dynamics of Innovation, Oxford University Press 1995
[11] Palass, B., Servatius, H.G., WissensWert – Mit Knowledge Management erfolgreich im E-Business, Schäffer-Poeschel 2001
[12] Meckel, M., Steinacker, L., Alles überall auf einmal – Wie Künstliche Intelligenz unsere Welt verändert und was wir dabei gewinnen können, Rowohlt 2024