AI That’s All Ears: The New Learning Coach

Since the coronavirus pandemic, more and more students have been suffering from mental health problems, but only 10% make use of existing counselling services. An interdisciplinary team at BFH is developing the “MindMate” app , which not only provides professional help, but also acts as a personal learning coach. We spoke to BFH researchers Lorenz Affolter and Daniel Reichenpfader as well as students Sara Lüthi and Jennifer Bürki, who created a functional prototype of the app as part of their bachelor’s thesis.

What was the trigger for the MindMate project?

Lorenz Affolter: Since the start of the coronavirus pandemic, there has been a massive increase in mental health problems among students. In response, the Swiss Student Union launched the Student Minds Network. At the same time, only around 10% of students in the canton of Bern make use of the existing counselling services – often due to a lack of knowledge, stigmatisation or barriers to access. This is where we want to start.

Daniel Reichenpfader: The good thing is that, technologically, we are now able to have very natural and helpful conversations using modern language models. Probably the best-known commercial tool for this is currently ChatGPT from OpenAI, Inc. However, there are also numerous freely available models and tools. We asked ourselves the following question: How can we use these technologies in a meaningful way to promote student mental health?

How did you, as students, get involved in this project?

Jennifer Bürki: We initially created a first version of the mobile app as part of a semester project. The topic really appealed to me personally, as I am very familiar with the challenges and pressure to perform during my studies. This then led to our bachelor’s thesis, in which we tested, expanded and improved the first prototype.

Sara Lüthi: I can only agree with that. What particularly fascinated me was the technical challenge: how do you get a chatbot to respond in a truly empathetic and helpful way? And how do you integrate this technology into a mobile app?

Can you explain the MindMate concept to us?

Lorenz Affolter: MindMate is more than just an ordinary chatbot. The system has two main components: The “Listener Bot”, which provides empathic support and directs you to professional counselling services when needed, and the “Learning Coach”, which helps you organise your studies.

Daniel Reichenpfader: Technically, the system is based on state-of-the-art large language models, which we integrate via a secure interface. The challenge was to configure the models in such a way that they can respond both empathetically and competently. We drew on the latest research findings to define the personality of the chatbot and support users in implementing behavioural changes.

What technical hurdles did you have to overcome?

Daniel Reichenpfader: The biggest challenge was prioritising the various requirements that were collected. We also work with sensitive mental health data, which requires the highest security standards. At the same time, the app should be low-threshold and easy to use.

Jennifer Bürki: It was particularly tricky to programme the chatbot in such a way that it could correctly assess crisis situations and connect users directly to an emergency telephone (e.g. the psychiatric emergency service at Inselspital Bern) in the event of an acute crisis. To do this, we used prompt engineering, created sample texts and evaluated the chatbot’s responses together with two psychologists.

How do you ensure that the app really meets the needs of the students?

Lorenz Affolter: We use participatory design methods. This means that the various stakeholders are involved in the development process right from the start. Of course, this primarily includes the students, but also student representatives, counsellors, experts and lecturers. We work closely with the Students Mind Network and Mindbalance organisations to ensure that we meet real needs.

Sara Lüthi: We carried out a usability test with nine students from different subject areas. Afterwards, they gave feedback on their impression of the app and formulated suggestions for improvement. It turned out that many of them were not even aware of the counselling services. We therefore integrated an overview directly into the app.

Where do you see the limits of such a system?

Daniel Reichenpfader: A chatbot can never replace professional psychological counselling. Our aim is to break down inhibitions and act as a bridge. MindMate should recognise when professional help is needed and then pass it on accordingly.

Lorenz Affolter: Exactly. We see MindMate as a preventative measure and a door opener. If we can get more students to seek support at an early stage, we will have already gained a lot. And by providing support with learning planning, we aim to make the app more attractive for all students and create added value and loyalty to the app even in non-crisis moments.

What’s next for the project?

Daniel Reichenpfader: We are currently preparing an application for further funding for the project. The prototype is currently running with a commercial language model, which we want to replace with our own local model as soon as possible. In terms of operation, we would like to integrate MindMate into existing communication platforms – messenger services, Moodle, campus app, etc. – in the long term. The more low-threshold and diverse the channels, the better.

Sara Lüthi: Our bachelor’s thesis will be completed in the summer. However, as we believe that the topic is extremely important and the current app provides a good basis for further development, we are motivated to continue our involvement. We already have lots of ideas for the next development phase, such as push notifications, more detailed views of the recorded tasks and improved, evidence-based learning planning.

What is your vision for the future of digital student support?

Lorenz Affolter: We see MindMate as a piece of the puzzle in student healthcare. As part of a related project, we are currently creating the basis for establishing a holistic student health management system at BFH. The MindMate app would fit in perfectly here.

Daniel Reichenpfader: Technology is developing rapidly. I see great opportunities in using AI-based systems responsibly for people’s well-being. MindMate is an important step in this direction, but we must continue to consider the risks of chatbots and try to minimise them.

Sara Lüthi: As a student, I hope that such tools will soon become standard. We need solutions that meet our generation where they are – digitally, holistically and without prejudice.

Jennifer Bürki: Ultimately, it’s about students no longer having to face their problems alone and knowing where they can go for help. If MindMate helps even just one person to find support in time, all the work will have been worth it.

Links

• Opportunities, Requirements, and Risks of a Student Mental Health Chatbot: A Qualitative User-Centred, Multi-Method Approach(https://ebooks.iospress.nl/volumearticle/72819)
• Resolution “Mental Health of Students” of the Association of Swiss Student Organisations(https://vss-unes.ch/wp-content/uploads/2023/05/2023-05-14_Resolution_psychGesundheit.pdf )
• BFH Student Health Management Project(https://www.bfh.ch/de/forschung/forschungsprojekte/2025-745-902-829/ )
• Article: Is This Chatbot Safe and Evidence-Based? A Call for the Critical Evaluation of Generative AI Mental Health Chatbots(https://jopm.jmir.org/2025/1/e69534 )
• Article: The FAIIR conversational AI agent assistant for youth mental health service provision(https://www.nature.com/articles/s41746-025-01647-6 )