The introduction of artificial intelligence (AI) offers many opportunities, but also presents companies with numerous challenges. Many companies have already implemented generative AI applications such as ChatGPT, but analytical methods including machine learning are also highly relevant. Many companies are still acting with caution when it comes to specialised use cases. The lack of courage to hand over competencies and tasks to AI is often due to uncertainties regarding costs and the risk that investments may not pay off immediately and that the chosen technologies may not be the right ones. A lack of transparency and traceability can also pose a risk. However, it does take courage to introduce AI. Fortunately, this courage can be translated into processes that set the guidelines for targeted use.

Overall, the implementation of AI requires a differentiated view of the opportunities and risks. We have compiled seven golden rules for implementing AI:

1. Develop an AI strategy

Before using AI, it is important to be clear about your goals. Take an unbiased approach and analyse the available options. It is crucial to regularly review functionality and results depending on the specific use case. A structured approach and comprehensive documentation are essential for development and implementation, as well as for subsequent transparency of use. An AI strategy that is consistent with the data and digital strategy and supports the corporate strategy ensures clear prioritisation. You should also bear in mind that continuous adjustments and improvements to AI systems are necessary to secure long-term competitive advantages.

2. Consider regulatory requirements

There are a number of legal requirements to consider, such as the EU General Data Protection Regulation (GDPR), which applies to the processing of personal data. For applications in the field of accounting, commercial and tax regulations must also be taken into account. In addition, the European AI Regulation, also known as the EU AI Act, came into force in 2024. The regulation governs the development, use and operation of AI systems in the EU. Germany must transpose the AI Act into national law by August 2026. Many applications are also subject to indirect transparency requirements, which demand that the development and functioning of AI be explainable, including adequate documentation and traceability.

3. Involve relevant stakeholders

Acceptance within the company is crucial for the successful implementation of AI. Employees play a central role in this and should be involved in projects at an early stage. Comprehensive information can help to transform potential reservations and uncertainties into confidence. Relevant stakeholders within the company should be continuously involved, as AI projects are not usually initiatives of the IT department. The success of such projects often depends on the cooperation between different disciplines.

4. Ensure data quality and security

Ensure adequate data quality in the underlying systems, as only valid data can lead to valid results. Traceability is also crucial. Otherwise, the AI system remains a ‘black box’ that cannot be trusted. Another important aspect is IT security, which refers to established protection goals such as confidentiality, integrity and availability. Authorisation, authenticity and bindingness are also of great importance. These aspects must be ensured during development as well as during application and in the technical operating environment.

5. Ensure effective governance processes

Most companies do not limit themselves to a single AI application, but use different solutions in different departments. It is therefore important to create a framework in which AI systems are developed and operated. If the processes are well structured, risks can be controlled more easily and the functioning of AI can be presented transparently. Through forward planning, these processes can be effectively integrated into existing management or internal control systems, which are often already in place. The better AI is integrated into daily operations, the less effort is required to meet the various requirements. Effectiveness can be demonstrated through both internal controls and external audits.

6. Promote training and further education measures

Digital literacy is not only important for developers, but also for users. It is important to develop a basic understanding of data structures, data sources and data quality, as well as to have adequate knowledge of the opportunities and, in particular, the risks of AI. This requires more than just basic user knowledge; it is necessary to have an understanding of how AI, machine learning and generative language models work. This is not only a requirement of Article 4 of the EU AI Act, but it is also crucial for assessing risks, developing effective prompting techniques and using AI appropriately.

7. Keep an eye on risks and requirements

Currently, AI applications are generally designed as assistance systems. It is important to consider the limitations of AI in a business context so that essential decisions ultimately remain with humans. Familiarise yourself with the risks as well as the functional and non-functional requirements at an early stage and take them into account in advance. This also includes ethical and legal requirements that go beyond the existing legal requirements. Another important aspect is the performance of the AI system, particularly with regard to the extent to which it can meet the requirements defined by the company in terms of functionality and results.

Artificial intelligence has become an indispensable part of the corporate landscape. Courage in dealing with AI means actively recognising and exploiting the opportunities offered by this technology, while at the same time managing the risks and ensuring responsible implementation.

Analytical AI will not only be able to detect anomalies, but also cluster these results and provide explanations. Technologies such as AI agents and deep research enable AI to take on increasingly complex tasks and, to a certain extent, perform its own quality assurance, which opens up further areas of application. By reducing fears and promoting a positive attitude towards AI, companies can drive innovation and gain competitive advantages. In this context, the (coordinated) use of AI is not only taken for granted, but is also considered a decisive factor for success.

This article was written by