Rödl AI Hub: The use of generative AI in auditing

The digitization of business processes in companies is steadily increasing. Expanding such digitization initiatives in companies with artificial intelligence (AI) systems no longer appears to be merely an option, but rather a necessity for gaining decisive competitive advantages. Generative AI in particular is important for transforming business models by enabling the implementation of autonomously acting AI agents. It is therefore essential for strategic decision-makers to understand the advantages associated with generative AI and its implementation possibilities and to use them for their own purposes.

Rödl & Partner realizes the benefits of AI through its in-house, self-developed Rödl AI Hub, which is used to implement numerous AI applications. This Rödl AI Hub is the result of collaboration between our Centre of Excellence, under the operational leadership of Edin Visnjic and David Klemme, and the respective digitalization managers from the individual business areas.

The Rödl AI Hub is a centralized, generic platform with a wide range of functionalities. Connection options for powerful generative AI models and essential basic functions, such as document vectorization, form the basis for valuable use cases in the field of auditing.

The flexibility of the Rödl AI Hub is particularly important to us, as it allows us to take into account the individual needs and requirements of our field of auditing. In addition, the Rödl AI Hub offers the possibility of connecting a wide variety of IT systems and databases, which facilitates the linking of all data sources with AI. This integration promotes comprehensive data analysis and utilization, which can significantly increase the efficiency and quality of informed decision-making processes. Furthermore, the integration of RPA & no-code/low-code solutions is possible for end-to-end process automation, with the Rödl AI Hub acting as the intelligent link.

When implementing use cases, Rödl & Partner relies on the concept of an AI-based co-pilot, in which our colleagues gradually expand their knowledge and methodological expertise in conjunction with AI. To implement the concept of an AI-based co-pilot in practice, AI cooperates with human experts to support a wide range of complex tasks. Both AI and human experts play to their strengths to optimize efficiency and effectiveness. Human experts contribute their methodological expertise, specialist knowledge and experience in specific subject areas to solve problems. The AI model, on the other hand, analyses large amounts of data, recognizes patterns and provides an initial indication of an interpretation. In addition, the human expert improves the AI system through regular feedback and at the same time benefits from the outputs generated by the AI system.

One approach to implementing the AI-based co-pilot within the Rödl AI Hub can be represented by specialized expert bots. These can access extensive databases validated by us at Rödl & Partner to generate reliable outputs with reference to the original documents using retrieval-augmented generation (RAG). The known weaknesses of generative AI, such as hallucinations, can be reduced to a minimum. Such outputs can be used to deal with complex problems. In addition, human experts can always rely on structured knowledge management for the performance of our audit and consulting services. This means that the Rödl AI Hub can also access licensed content such as auditing standards, accounting standards or the WP Handbook when generating answers based on complex questions. This enables us to output generative AI with valuable and reliable source references and serves as helpful support for our human experts in solving complex problems. This ability to access licensed content is the result of cooperation with the relevant publisher, in our specific example with IDW Verlag.

Another advantage of our Rödl AI Hub is the ability to implement different protection classes for underlying AI models depending on the type of data to be used. This allows us to specify, among other things, the use cases for which AI models are to be deployed on in-house servers or solutions from cloud providers. This flexibility enables us at Rödl & Partner to fully comply with regulatory requirements while maximizing the performance of AI.

In addition, the Rödl AI Hub allows the implementation of the latest generative AI functionalities by enabling agent-based workflows. In such workflows, each agent takes on a specific subtask to solve the overall task. The workflow is usually controlled by a central orchestrator, which autonomously controls the individual agents and receives their output within the workflow, in order to then process it for an overall statement. This structure optimizes the efficiency and responsiveness of business processes by enabling the automation and coordination of complex tasks through agent-based workflows, thus representing the next level of AI application in companies.

With all this information contained in the Rödl AI Hub and the associated cooperation between humans and machines, we at Rödl & Partner can take our service portfolio to a new level. In addition, we are always there to support you with any questions or challenges you may have in connection with the introduction of AI into your business processes. We are happy to help you determine the right technologies, optimize your processes and ensure that they comply with legal requirements.