Dear colleagues,
As the date approaches, we would like to kindly invite you to the upcoming Constructor Talk, jointly organised by Constructor Knowledge Labs (CKL) and the School of Computer Science and Engineering (CSE).
This Talk will examine how agentic AI is fundamentally changing the economics and structure of scientific work, enabling faster iteration, deeper validation, and more ambitious research trajectories.
Date & Time: 22 April, 13:00-14:00 CET
Format: On-site and online
Location: IRC Conference Room
Join online: https://lnkd.in/ghNJj_5k
Meeting ID: 991 3392 6066, Passcode: 416988
Light catering and refreshments will be available on-site, and we warmly encourage colleagues to join us in person for informal discussion and networking following the seminar.
Speaker: Prof. Dr. Andrey Ustyuzhanin
Principal Investigator at Constructor Knowledge Labs, Adjunct Professor of Computer Science at Constructor University Bremen.
Talk: Agentic Flows for Scientific Research: A Practical Perspective
In this Talk, Andrey Ustyuzhanin shares his experience of introducing agentic AI into the daily workflow of a university research lab, working with tools such as Claude Code, multi-agent pipelines, and structured knowledge management.
He starts from a simple principle: before spending GPU cycles, spend tokens. Each hypothesis is first subjected to automated, multi-layered critique across logic, methodology, and argumentation, before any experimental work begins. Research is then organized as a sequence of phased steps with clear go/no-go decisions, where responsibilities are distributed across agent roles such as planner, engineer, reviewer, runner, and analyst, reflecting how well-functioning research teams operate, but with much shorter iteration cycles.
Drawing on concrete examples, he also addresses the limitations of this approach: hallucinations that make unsupervised outputs unreliable, gaps in domain knowledge that lead to implausible results, and the open question of how to report and reproduce agent-assisted research.
The talk focuses on how this way of working shortens the path from idea to evidence, making it easier to pursue exploratory, high-risk hypotheses, while bringing more structure and discipline into the research process.
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
If you are dealing with AI in your work or thinking about how research is changing in practice, we warmly welcome you to join us. CKL aims to keep the session open and discussion-driven, with space for different perspectives and honest exchange.
Warm regards,
Constructor Knowledge Labs
CKL Continues Its Monthly Series with Constructor Talk: Agentic Flows for Scientific Research: A Practical Perspective (22 April)
