Constructor Knowledge Labs (CKL), in collaboration with the School of Computer Science and Engineering (CSE), continues its successful seminar series with its third session, to be held on 22 April, 13:00–14:00 CET.
This upcoming seminar will focus on the transformative role of agentic AI systems in scientific research workflows, with particular attention to how such systems reshape the economics, structure, and pace of research. The session will explore how agent-based approaches enable faster iteration, more rigorous validation, and the pursuit of more ambitious and exploratory research directions.
The keynote will be delivered by Andrey Ustyuzhanin, Principal Investigator at Constructor Knowledge Labs and Adjunct Professor of Computer Science at Constructor University Bremen.
Talk Title: Agentic Flows for Scientific Research: A Practical Perspective
In this talk, Prof. Ustyuzhanin will present practical insights from integrating agentic AI into the daily operations of a university research lab. The approach is grounded in a key principle: prioritizing structured reasoning and critique before computational experimentation. Research hypotheses are first evaluated through automated, multi-layered analysis – covering logic, methodology, and argumentation -before allocating GPU resources.
The workflow is organized into clearly defined phases with explicit go/no-go decision points, distributing responsibilities across specialized agent roles such as planner, engineer, reviewer, runner, and analyst. This structure mirrors high-functioning research teams while significantly accelerating iteration cycles.
Drawing on real-world examples, the talk will also address current limitations of agentic systems, including hallucinations, gaps in domain knowledge, and challenges related to reproducibility and reporting in agent-assisted research.
Overall, the session highlights how agentic workflows can shorten the path from hypothesis to evidence, enabling more efficient exploration of high-risk, high-reward ideas while introducing greater structure and discipline into the research process.
Event Details
- Date & Time: 22 April 2026, 13:00–14:00 CET
- Format: On-site and online
- Location: IRC Conference Room
- Online Participation: https://lnkd.in/ghNJj_5k
- Meeting ID: 991 3392 6066, Passcode: 416988
Agenda
- 13:00–13:05 – Introduction & Opening
- 13:05–13:50 – Keynote Presentation by Prof. Dr. Andrey Ustyuzhanin
- 13:50–14:00 – Discussion and Q&A Session followed by informal discussion over coffee
Researchers, practitioners, and anyone interested in the evolving role of AI in scientific practice are warmly invited to attend. The CKL seminar series is designed to foster open, discussion-driven exchange across disciplines.