Relevant AI industry projects

In cooperation with an energy supply company, we are developing a large language model to generate standardized requirements from text information. This standardization reduces the time needed to revise existing requirement information and enables the creation of future requirements in a consistent manner. This is significant for project management, affecting risks, quality, and costs. The process involves collecting and preparing data, implementing the model, and training it. The result is a demonstrator capable of performing the task in an experimental setting. This development is an important step in the automation of requirements engineering.

In collaboration with a leading energy company, organization-specific AI use cases in ongoing projects were identified and developed through multiple development cycles. Various factors including organizational, domain-specific and technological contexts were taken into account. Our extensive support ranged from rapid prototyping to analyzing organizational impacts and assessing implications for change management. Throughout this process, we assisted the company in conducting AI experiments, facilitating the testing of feasibility and effectiveness in actual operational environments. The AI use cases were subsequently prepared, consolidated, and evaluated to provide valuable insights for decision-making.

In our co-branded study with Celonis, a global leader in process mining, we investigated the interplay between Process Mining and generative AI. In total, we conducted interviews with 14 Process Mining thought leaders from the CeloCoE community and academia. As scientific groundwork, we employed a multifaceted approach that involved an extensive review of existing research and the integration of our own contributions in the field. Thereby, we found that both technologies substantially benefit from each other: while process mining offers a reliable data backbone to support the uptake of generative AI at the enterprise level, generative AI enhances existing and enables new process mining capabilities. Therefore, leaders of Process Mining Centers of Excellence need to take action now, with our research providing actionable practices that will serve as a foundation to prepare for the transformative journey ahead with generative AI.

The continuous development of artificial intelligence increasingly shapes both our personal and professional lives and leading to increased interactions with AI systems more and more often. In collaboration with the Ernst & Young GmbH Wirtschaftsprüfungsgesellschaft, we therefore analyzed how companies can successfully establish interactions with AI. Based on extensive scientific research as well as interviews with experts and providers of AI solutions, five different types of interaction have been identified, each distinguishable by its characteristic features. From these findings, we derived implications and recommendations for companies, society, and politics. Our focus was not only on the successful design of current application scenarios, but also on the potential of future human-AI interaction. In this context, ten theses were formulated which summarize the essential changes in human-AI interactions. Subsequently, the key results from this research were published in a collaborative publication.

In this project, the goal was to develop and implement a decision support system for efficient troubleshooting of printer error messages to increase service quality and improve resource efficiency. The primary goal was to prevent additional costs arising from extended printer downtimes and minimize unnecessary technician visits. Therefore, we created a proof of concept (PoC) in the form of an AI prototype demonstrating the potential of our developed machine learning approach. The AI prototype allows for text- and error code-based identification of machine spare parts to increase the first-fix rate while maintaining service quality. Our project included capturing and analyzing the current situation, supporting the selection of suitable software as well as AI algorithms and implementing appropriate models. Additionally, a comprehensive onboarding strategy for development partners was developed to ensure a seamless integration of the solutions into existing structures.

The focus of this project was on a detailed analysis of resource consumption in the manufacturing of food products in close collaboration with a leading food production company. At the core of our efforts, the development of an AI-based approach to capture the economic savings potential through an resource-efficient use of resources was undertaken. The challenges were addressed by the creation of a clear overview of the economic potential and the identification of specific measures for implementing savings using AI. With the ending of the project, the results were visually presented, and an interactive management dashboard for resource consumption was developed. The objective was to promote the sustainability of production practices, including the documentation and integration of the dashboard into the IT infrastructure of the production site.