Aarhus Agentic AI Exchange
Building AI Agents: Diversity, Safety, and Practice
Info about event
Time
Location
AIAS, building 1632, room 212
Join researchers and builders from across Aarhus to explore agentic AI through three perspectives: diversity, safety, and practical experimentation. Short talks and open discussion will connect research questions with building experience, followed by pizza and networking.
Intended outcomes
- Establish a local network of researchers and builders who meet regularly to share work, exchange ideas, and collaborate.
- Understand key challenges around diversity, safety, and security in agentic systems.
- Identify practical starting points for experimenting with AI applications and agentic workflows using UCloud AI.
- Gather shared questions, potential collaborations, and topics for future meetups
Program
The speaker order and timings below are provisional and can be adjusted once the contributions are agreed.
| Time | Session | Speaker/facilitator | Intended outcome |
| 15:00-15:10 | Welcome, introductions, and framing | Kristoffer Nielbo and Andreas Roepstorff | Establish the central question and participants' interests |
| 15:10-15:35 | The Diversity Challenge in Agentic Systems | Tianyi (Tenney) Hu | Key idea, finding, or example |
| 15:35-16:00 | A Graphical Foundation for Agentic Safety and Security | Nicolas Legrand | Question to explore or perspectives to compare |
| 16:00-16:10 | Break and informal exchange | - | Connect and regroup |
| 16:10-16:45 | UCloud AI for builders | Emiliano Molinaro | Concrete output or shared learning |
| 16:45-17:00 | Takeaways, open questions, and next steps | Kristoffer Nielbo and Andreas Roepstorff | Capture useful findings and possible follow-ups |
| 17:00 onwards | Pizza and networking | - | Informal conversation and new connections |
Speakers
Tianyi (Tenney) Hu, CLAN Lab, Computer Science, Aarhus University
Title and focus: The Diversity Challenge in Agentic Systems
Description: Agentic large language models are reshaping how people understand the world and make decisions, raising fundamental questions about the diversity of perspectives they represent and the biases they may reinforce. We will explore these challenges and potential directions for building systems that support diverse perspectives and informed decision-making.
Nicolas Legrand, Center for Humanities Computing, Arts, Aarhus University
Title and focus: A Graphical Foundation for Agentic Safety and Security
Description: AI systems are rapidly shifting from passive models to autonomous agents that plan, act, collaborate, and decide across complex workflows, introducing new classes of safety and cybersecurity risks. We develop a graphical foundation for modeling the dependencies and decision chains of agentic systems, enabling systematic analysis of risk propagation under dynamic threats and environments.
Emiliano Molinaro, SDU eScience Center, University of Southern Denmark
Title and focus: UCloud AI for Builders
Description: A practical introduction to UCloud AI for researchers and developers exploring agentic workflows. The talk will discuss getting started, supporting experimentation, and the practical challenges of turning ideas into working AI applications.
Participants and prior knowledge
The meetup welcomes researchers, students, developers, practitioners, and other interested participants across disciplines, sectors, and career stages. An interest in agentic AI is the only prerequisite; no programming experience or specialist background is required.
Due to catering all participants must register. Please sign up via the button above or here. Limited seating, first come, first serve.
Contact
Kamilla Rosenberg Franck, kamilla@aias.au.dk