On December 11, 2024, the final presentation of the FPV project titled "GenAI in the Construction Industry" took place with Cellere Bau AG as the industry partner.

The aim of the project was to optimize the tendering process in the construction industry through the targeted use of AI tools. Two key challenges were addressed: identifying a suitable AI model and developing an approach to reduce the manual effort required for analyzing tender documents. The project team, consisting of Alexa Meier, Jonas Jaeggi, Nico Lamprecht, and Phillip Kessel, worked under the supervision of Prof. Dr. Andrea Back, Maximilian Breitruck (HSG), and Kai Blanke (Cellere Bau AG).

By conducting a thorough market analysis and extensive stress tests of various tools, the team developed a practical solution that not only accelerates the tendering process but also significantly improves its efficiency. However, the project also revealed some limitations in integrating GenAI into existing workflows. Current solutions can automate specific steps effectively but are often unable to comprehensively handle complex processes. Additionally, interactions between humans and AI models still require clear instructions and structured training to achieve optimal results and fully leverage the benefits of the technology.

Approach and Results

1. Market Analysis and Tool Evaluation: Analysis of available AI models (e.g., ChatGPT, Copilot) based on functionality, cost, and integration capabilities.
2. Tool Testing and Stress Tests: Application of real tender data to evaluate performance, leading to the development of a custom GPT tailored to Cellere’s specific requirements.
3. Automation of Manual Processes: The analysis of tender documents was automated, resulting in significant time savings and reduced errors.
4. Development of Training and Implementation Concepts: Structured guidelines and prompt aids were created to prepare users for the practical application of the solution.

Practical Value

The automation allowed for a significant reallocation of work time and made the tendering process much more efficient. The project’s outcomes provide a scalable solution that can now be applied by the industry partner to other existing processes. At the same time, technical limitations were identified and specifically addressed during the project.

A big thank you to Kai Blanke and the team for the excellent collaboration!