Either traditional or agile software project, the role of the managers is very hard. One of the toughest tasks is communication of the key product values to be implemented (i.e. requirements) with customers and other stakeholders and presenting them to the development team. We consider this role as a separate one - these human proxies between the business and development are the product owners (PO). In an educational setup, where the students' task is any development task, the teacher is playing the role of the PO, reflecting various requirements to the students. Since the student teams are different, this reflection should also vary if possible. In such a varying environment, even more after considering the need of uniqueness of assignments every semester, the workload for the teacher is increasing. A common way to deal with this problem is using pre-scheduled presentation of constant task details, e.g. in selected weeks of the semester. Other way of minimizing growth of the workload is using simulators, where the students get their task details not pre-scheduled, but based on the proofs of their prior achievements, e.g. by submitting some files to a document management system. We can find a possible use of artificial intelligence (AI) in this problem field as well. In our presented approach, generative AI is used in artificial PO implementation in a particularly simplified situation. We integrate existing generative AI solution to produce new and modified requirements based on a prior software application domain definition, travelling app, in our case. These requirements get stored in a database making the event based scheduling of their appearance possible. To move part of the learning and task solving out of the classroom, the user interface to the artificial PO is implemented as a multi-platform mobile application (using Flutter technology). Even that our aim is to simulate an agile project, in this paper we mainly focus on the requirement generating artificial PO application. Our evaluation focuses on the abilities of the artificial PO, without waiting for the development tasks to finish. The presented results show usability in several educational subjects in software engineering that require frequent, domain specific consultations of product or implementation details.

Keywords: Generative artificial intelligence, agile project, project owner, Flutter.