Ingo Eichhorst
High Speed Open Source: Ich wollte nur ein kleines Tool bauen, jetzt leite ich eine Softwarefabrik · How coding agents can be directed, controlled and sustained in fast-paced open-source work.
Programme insights
A look back at themes, questions and perspectives from the Berlin Expert Days 2026 programme.
This selection shows programme items from AI & Agentic Engineering and software architecture. It complements the full schedule without ranking other sessions.
AI between speed, operations and responsibility
These three sessions made one point clear: productive use of AI needs more than a convincing model. It calls for robust system boundaries, traceable quality, human control and a conscious approach to its effects on our work.
Michael Seel
AI Systems Architecture: What makes AI features production-ready through clear interfaces, evaluation, policies and observability.
Martin Lippert
Cognitive Debt – Was KI mit unseren Köpfen macht · What happens when we delegate programming work to AI, and how to retain our ability to review it.
Making knowledge available, understanding systems
From a chatbot’s memory to the analysis of long-lived system landscapes: these perspectives focused on context, sources and tools that turn information into reliable insight.
Ivan Vokhmin
From RAG to memory decay: Techniques such as reranking and controlled forgetting for helpful, fast AI contexts.
Gerd Aschemann
The Excel Moment: AI for Legacy Architecture Analysis · How LLM-based tools bring together code, tickets and documentation to analyse legacy systems.
Building agents, understanding risks, involving people
Agentic systems change technical possibilities and responsibilities at the same time. These sessions connected Spring AI, secure systems and collaborative methods with the question of where human expertise remains indispensable.
Christoph Menzel
AI: Game Changer or Game Over? - The Security Paradox of Artificial Intelligence · The opportunities for AI in security and the new attack surface of autonomous systems.
Carsten Lill
Kollaborative Methoden und KI: Wie „Human in the Loop“ Spaß machen kann · How teams can use collaborative modelling and AI without relinquishing control of their decisions.
Timo Salm
Building AI Agents with Spring AI · From tool calls and orchestration to reusable AI skills on the JVM.
Architectural work under new conditions
AI does not make architectural questions smaller: it shifts the focus to clarity, system understanding and the development of skills. Two sessions explored these questions from first principles to future professional roles.
Uwe Friedrichsen
Essential architectural thinking - why, what, how, when and how much? · What architectural work does, when it is needed and how AI can support it.
Alexander Lehmann
Wir brauchen keine Entwickler mehr, wir brauchen Solution Architects · Why problem analysis, system structuring and architectural skills gain importance in the age of generative AI.
Architecture as a shared practice
Architecture does not emerge only in documents. These sessions explored decisions, evolutionary practices and a productive approach to customer-specific requirements – close to teams’ daily work.
Falk Sippach
Emergente Praktiken in der Softwarearchitektur · How ADRs, fitness functions and collaborative methods provide direction without rigidly prescribing architecture.
Gerrit Beine
Ihr entscheidet alle immer falsch! · A different view of architectural decisions, cognitive biases and the tools for them.
Tobias Larscheid
Customization pro Kunde architektonisch verankern statt ständig nach Standardisierung zu rufen · How product architecture can support individual requirements for many customers with reasonable effort.
Sponsors
Partners
