DIGITAL LIBRARY
DESIGNING A MULTI-STAKEHOLDER THESIS SUPPORT ASSISTANT FOR INFORMATION TECHNOLOGY BACHELOR PROGRAMMES
Tallinn University of Technology (ESTONIA)
About this paper:
Appears in: EDULEARN26 Proceedings
Publication year: 2026
Article: 1604
ISBN: 978-84-09-88444-5
ISSN: 2340-1117
doi: 10.21125/edulearn.2026.1604
Conference name: 18th International Conference on Education and New Learning Technologies
Dates: 29 June-1 July, 2026
Location: Palma, Spain
Abstract:
Bachelor thesis completion can be challenging for many students in higher education, especially in the final study year when they must combine independent research, writing, supervision meetings, administrative requirements, and defence preparation. This paper presents the early design phase of a one-year educational development programme that aims to support thesis processes in six information technology bachelor programmes through a multi-stakeholder assistant based on retrieval-augmented generation (RAG). The planned assistant is intended to provide grounded answers from institutional sources to six main stakeholder groups: students, supervisors, programme leaders, reviewers, administrative staff, and defence committee members.

The study followed a sequential design approach. First, six individual interviews were conducted with programme leaders to identify the main bottlenecks and recurring failure points in current thesis workflows. Second, a two-hour Design Thinking workshop with 25 academic and administrative staff was organised to map the questions that the future assistant should answer for each stakeholder group. The next stages include collecting and structuring institutional materials for the knowledge base, developing the RAG assistant, and preparing stakeholder onboarding. A follow-up workshop in May 2026 will focus on how to introduce the assistant so that it is actually adopted in everyday practice.

The initial results indicate that the key challenge is not only academic writing itself but the fragmentation of the overall process. Participants highlighted uncertainty related to topic selection, supervisor-student alignment, milestones, formal requirements, assessment expectations, and communication responsibilities. The workshop resulted in a structured question map that now serves as the basis for knowledge base construction and prompt design. The project will continue with technical development and user testing, followed by a 2026/2027 pilot with at least 200 students and 30 teachers across a minimum of three programmes.

The paper contributes a practical model for designing a process-oriented thesis support assistant in higher education. Rather than replacing academic supervision, the proposed solution aims to make institutional knowledge more accessible, reduce avoidable delays, and support more consistent guidance across stakeholder groups.
Keywords:
Retrieval-augmented generation, bachelor thesis, higher education, student support, academic workflows, information technology education.