BeeChat
The project designed, developed and evaluated a data-privacy-compliant, locally operated AI infrastructure for Bern University of Applied Sciences.
Factsheet
- Schools involved School of Engineering and Computer Science
- Institute(s) Institute for Patient-centered Digital Health (PCDH)
- Research unit(s) PCDH / AI for Health
- Funding organisation BFH
- Duration 01.01.2024 - 31.12.2025
- Head of project Prof. Dr. Kerstin Denecke
Situation
We began by investigating the requirements of students, lecturers and staff regarding generative AI use in the higher education context. On this basis, we used open-source technologies to build a modular platform that supports various language models, additional functions and integrations. In parallel, we developed methods to evaluate response quality, resource utilisation and scalability to creating the necessary capabilities for wider institutional adoption. As we gained further insights, we incorporated them iteratively into the ongoing development of the technical platform.
Result
The project initially culminated in BeeChat, a functional prototype for the data-privacy-compliant deployment of generative AI within higher education. The development phase validated the technical feasibility of a on-premises AI platform and provided valuable insights into its operation, governance, scalability and user requirements. Building on these outcomes, the project established the underlying technical framework for the current TI MLMP inference service, which facilitates the centralised provisioning and management of open-source language models across BFH. This has created a sustainable infrastructure that supports diverse research, teaching and innovation projects and provides an independent alternative to commercial AI services.
Looking ahead
The foundations laid by this project provide the basis for expanding the TI MLMP inference service into BFH’s central AI infrastructure. Future development will focus on integrating additional models and AI features, adding support for other research and teaching applications and continuously improving scalability, security and usability. Ultimately, the platform aims to support the responsible use of generative AI, enable innovative applications in teaching and research, and strengthen BFH’s technological independence in the field of AI.
