🎥AI as Co-Creator – Developing AI-Supported Learning Bots for Teaching, Guidance and Practice Scenarios

Description

Julian van der Kraats, product owner in the IT department of Leiden University, is developing LUCA (Leiden University Cognitive Agent), a platform that enables teachers to easily create AI-supported learning bots. Using templates ranging from Socratic tutors to practice bots, educators can design, deploy, and integrate their own AI agents into their courses without any technical expertise.

The aim of the project is to lower the threshold for teachers to work with AI, support practice-oriented teaching scenarios, and systematically gather experience with AI in higher education.

Initial Situation

Many instructors want to integrate AI into their teaching but struggle with technical barriers or time constraints. At the same time, students increasingly seek flexible support in understanding difficult concepts, solving tasks, or clarifying administrative questions. LUCA addresses this challenge by providing a secure, user-friendly platform for creating and deploying educational AI bots.

“We noticed that teachers find it very hard to get started with AI.” – Julian van der Kraats

Ziele

• Enable teachers to design AI learning bots without technical knowledge
• Improve students’ conceptual understanding and increase opportunities for practice
• Provide realistic role-play simulations for skill development
• Ensure transparency and data protection
• Collect evidence on the educational impact of AI-supported bots

“Right now it’s very important to gain experience on how well AI works in an educational setting.” – Julian van der Kraats

Concepts, Implementation, Methods

LUCA offers modular templates that allow teachers to build AI bots for different scenarios: Socratic dialogue, question-answering functions, or role-play simulations. An integrated assistant agent supports educators in structuring content and refining prompts. Students use these bots intensively — not only for subject-related questions but also for organizational matters. Since each bot is connected to course materials, it provides precise and context-aware answers. A standout example is a role-play bot used by master’s students in cognitive psychology to practice assessing suicide risk. The bot realistically simulates a person with suicidal tendencies and only reveals critical information when students ask targeted diagnostic questions. This trains communication and risk assessment skills in a safe environment. Another bot guides researchers step by step through creating a data management plan and automatically generates a draft for submission.

Experiences

• Students ask a wide range of questions, and bots reliably respond with clarity and patience.
• Complex concepts become more accessible because the bot never “rushes” explanations.
• Role-play bots support the development of diagnostic and communicative competencies.
• The data-plan bot significantly reduces researchers’ administrative workload.
• Transparent communication about data protection increases acceptance.

“Students ask all kinds of questions and the bot patiently explains the concepts every time.” – Julian van der Kraats

Success Criteria

• Teachers can build functional AI bots without technical expertise.
• Students receive consistent and understandable support during learning and practice.
• Role-play bots enable realistic training of complex communication skills.
• Researchers benefit from reduced administrative burden.
• The system provides transparent, reliable data protection.

“The data is as safe as your files in OneDrive and nothing is used to train any model.” – Julian van der Kraats