We proudly present our brand new TUM MGT chatbot, pAIge!
The research project "Generative AI in the School Office of the TUM School of Management - Use Case TUM MGT Chatbot" has collaborated with several stakeholders from industry and academiato develop an innovative AI chatbot that provides answers to frequently asked questions, thereby freeing up capacity in the student support team.
Students and stakeholders such as the Technical University of Munich, the TUM School of Computation, Information and Technology, Microsoft, and the Bavarian State Institute for Higher Education Research and Planning (Bayerisches Staatsinstitut für Hochschulforschung und Hochschulplanung IHF) were heavily involved in the implementation, including an exciting hackathon at Microsoft headquarters to develop the MVP product on the Azure OpenAI platform. The collaboration with the IHF is part of a research experiment.
The use of the chatbot "pAIge" will not only reduce the consulting workload but also free up management resources.
At the same time, the TUM School of Management and Technical University of Munich position themselves as pioneers in the use of generativeAI and drive innovation in the educational landscape.
From students for students
The special thing about the technical assistant: It was developed by students for students. And who knows the concerns of young academics better than they do? How practical that two current TUM students in Management & Technology, Henri Zalbertus and Jan Plüer, have taken on the project management for the chatbot: "We met at the TUM Campus Heilbronn and got on well straight away due to our competitive sports backgrounds," says Zalbertus. The two now share an apartment in Munich, have founded their own start-up and are working on the chatbot project with their team.
Application letter via hackathon
The two 23-year-olds were approached by project initiator Barbara Tasch, Managing Director of the TUM School of Management. "In November, we organized a hackathon by and with Microsoft as a kind of application event." The task: develop a functioning chatbot for student concerns in three days. Synergies were also used with Stephan Krusche, Professor of Information Engineering at TUM Campus Heilbronn. He developed a chatbot for programming tasks called Iris.
The team was quickly formed and the budget was provided by TUM, so they could get started. "It was quite something to guide so many people at once and then in such a complex context," Henri looks back. The first task was to collect data. Zalbertus and Plüer had to find out what the students' typical problems were. "We conducted interviews and evaluated anonymized emails." Then they fed their database with administrative information from the website and documents from the download center. That can be quite stressful, can't it? "One of our greatest strengths is, if you ask others, rather our greatest weakness, that we are so homogeneous and basically never argue," says Zalbertus and laughs.
The starting line
As with all chatbots, the rule of thumb of machine learning applies: what goes in comes out. "The data is so fundamental that at the end you can see which documents need to be adjusted or deleted and whether the chatbot really provides helpful answers."
Henri sees the end of the project with one laughing and one crying eye: "Some people are very sad now. Our team of eight worked really well together, maybe we can get some of them to join our start-up." The success of the young managers and computer scientists shows what is possible with the right support. "Barbara Tasch placed a lot of trust in us, without her and the commitment of everyone involved, this would not have been possible."
Congratulations on the launch of the chatbot, dear project team: Barbara Tasch, Jan Plüer, Henri Zalbertus, Joshua Milbers, Benedikt Blümelhuber, Penny Hung, Michael Dyer, Mohanad Kandil, Leon Henrik Thiel, Thomas Wölkhart, Gentrit Fazlija!
And a big thank you to all stakeholders for their tremendous support:
Alexander Braun, Adil Boushib, Robert Eichenseer, Jürgen Ernstberger, Dr. Susanne Falk, Gunther Friedl, Georg Groh, Jochen Hartmann, Stephan Krusche, Lorenz Kupfer, Dr. Stefanie Lämmle, Patrick Melchner, Hana Milanov, David Podolskyi, Dr. Jasper Schwenzow, Michael Strobel, Daniela Todorova, Isabell Welpe.