AIDE#26
AI Assisted Design & Engineering of Digital Solutions & Systems

AI is transforming how we design and engineer digital solutions — but the real question isn't whether it's powerful. It's whether we're mature enough to use it well.

Fun: In just hours, structured templates carried Kim through many iterations between concept and code, resulting in a strong prototype ready for end-user tests. Genuinely exciting — but also humbling: AI costs can run high. The key? Having fun while staying critical.

Facts: Reliable research is still scarce, but early findings reveal a consistent pattern: balanced effort across the entire design and engineering lifecycle has always been the sweet spot. Accelerating just coding or just design creates bottlenecks elsewhere. End-to-end thinking matters more than ever.

Fiction: Many people expect AI to be the one tool that solves every problem. Kim's experiments show a different reality: AI amplifies what we give it — good ideas and bad ones alike. You must master both your AI and your methodology to judge quality.

Takeaway: AI doesn't replace expertise. It's a partner on equal footing. AI brings speed and scale; you bring judgment and methodological depth. The stronger your fundamentals, the more powerful the collaboration becomes.

Kim Lauenroth is Chairman of IREB e.V. and has been designing digital solutions since 2011. Since 2022, he has been teaching digital design at Dortmund University of Applied Sciences and Arts. As a co-initiator of the Digital Design Manifesto, he has been advocating for years for digital design as a distinct profession. Kim studied computer science, business administration, and psychology, received his Ph.D. in 2009 on requirements engineering for product lines, and has authored textbooks and more than 40 publications.

Artificial Intelligence is often seen as a game changer for Requirements Engineering, promising faster delivery, better quality, and less manual effort. But how much of this holds true in real project environments?

This session provides a practical perspective based on hands-on experience with different AI tools. It explores where AI can effectively support common Requirements Engineering tasks, such as structuring requirements, generating user stories in predefined formats, or identifying potential edge cases. In these areas, AI can help improve efficiency and consistency.

At the same time, the session highlights important limitations. Requirements Engineering is not only about documentation, but about understanding context, making decisions, and working with people. Stakeholder communication, individual expectations, and situational judgement remain critical and cannot be replaced by AI.

Another key aspect is the role of the practitioner. Using AI effectively requires domain knowledge, the ability to formulate meaningful prompts, and the competence to critically evaluate results. Without this, AI outputs can be misleading or incomplete.

This talk offers a realistic and balanced view on AI in Requirements Engineering: what actually works, what does not, and how it can be used as a practical support tool without overestimating its capabilities.

Leonie Hahn works in consulting, specializing in Requirements Engineering, product ownership, and digital transformation. She has led international development teams and worked on large-scale projects in the automotive and defense sector, with focus on cloud migration and application modernization. Her work centers on structuring complex requirements, aligning stakeholders, and ensuring successful delivery in agile environments. Through her hands-on experience, she brings a pragmatic perspective on how emerging technologies such as AI can support, but not replace, the core principles of Requirements Engineering.

Prototyping ist ein zentraler Teil des Designprozesses, bei dem Produktdesigner Ideen und Konzepte ausprobieren, verwerfen und neu entdecken. AI-Prototyping verspricht hier Beschleunigung und neue Möglichkeiten, wie wir die Probleme unserer Nutzer:innen besser verstehen und Lösungen finden können.

Aber was passiert mit all den Entscheidungen, die beim Prototyping gefällt werden, wenn AI die Ausführung übernimmt? Und wie vermeiden wir das Risiko, dass wir nur „Slop" erzeugen – also schnell generierte, ähnliche Prototypen mit geringem Lerneffekt?

Diese Session nimmt Erfahrungen aus der Designpraxis auf und zeigt, wo AI-Prototyping die Exploration beschleunigt, Ideenvielfalt fördert – und wo es den kreativen Prozess einschränkt.

Bio coming soon.

Architecture in the Age of AI-Agents — Why AI-Driven Software Engineering Requires AI-Consumable Systems

AI-assisted coding has dramatically accelerated software development. But it has also revealed a fundamental limitation: optimizing isolated engineering tasks with AI does not automatically create better systems. In many cases, it creates the opposite — inconsistency, fragmentation, and architectural decay. AI is changing what limits software engineering productivity. In the past, the main bottleneck was writing and implementing code. As code generation becomes increasingly automated, other factors become critical: the quality of the context provided to AI, the structure of the system, the clarity of the architecture, and the ability to orchestrate reliable engineering processes end-to-end. This keynote argues that the real transformation is therefore not about faster coding, but about a new production model for software engineering: agentic, factory-like systems in which AI agents operate within structured processes, guided by quality gates, feedback loops, architectural constraints, and human oversight. Similar to industrial production, value no longer emerges from isolated craftsmanship alone, but from designing reliable, scalable, and continuously improving production flows for software systems. At the center of this transformation lies a critical insight: architecture determines the effectiveness of AI-driven engineering. Modern AI systems do not understand software in the human sense — they operate on context. When systems are modular, semantically consistent, and architecturally well-structured, AI becomes precise and reliable. When they are not, increasing model capability merely amplifies noise, instability, and complexity. This fundamentally changes the role of architecture. Software systems are no longer designed only for human developers, but increasingly for machine agents that depend on navigable context, explicit boundaries, structured knowledge, and AI-consumable system designs. Building on real-world experience and architectural principles, this keynote introduces the concept of “AI-consumable architecture” as the foundation for scalable agentic engineering and software factories. Positioned within the broader evolution toward end-to-end AI maturity, the talk connects requirements engineering, architecture, implementation, and operational feedback into a unified perspective on AI-driven software development.

The takeaway is clear: AI will dramatically accelerate software engineering — but the organizations that succeed will not be the ones generating code the fastest. They will be the ones building architectures that allow AI agents to operate reliably, continuously, and at scale.

Dr. Sönke J. Magnussen is an IT project manager, consultant, and software architect at WPS – Workplace Solutions GmbH in Hamburg. He brings extensive experience in software architecture, agile engineering practices, and the practical application of AI in software development. As a member of iSAQB, he contributes to shaping the future of software architecture education and standards.

Code generation · Architectural decision support · Refactoring at scale · Technical debt management

Speaker to be confirmed.

Test generation · Defect prediction · Automated quality gates · Continuous testing

Speaker to be confirmed.

Moderator and 4–5 expert panelists representing IREB, iSAQB, and industry perspectives. Closing with audience Q&A.

Where are the biggest gaps in current AI tool ecosystems? · How do we prevent bottlenecks when one phase accelerates? · Bridging requirements, design, and architecture in AI-assisted workflows · Skills and roles: What changes with end-to-end AI maturity? · Governance and responsibility in AI-assisted workflows · Standards and best practices: The role of IREB and iSAQB in the AI era

Hands-on workshop · Max. 30 participants · IREB-led
Facilitator to be confirmed.

AI-powered requirements workshops and elicitation · Design system generation with AI tools · Prompt engineering for designers and requirements engineers · AI-assisted user research and prototyping

Hands-on workshop · Max. 30 participants · iSAQB-led
Facilitator to be confirmed.

AI-assisted architectural decision-making · Code generation and review workflows · AI-powered refactoring and technical debt management · Architecture documentation with LLMs

Both tutorial tracks reconvene in the main room for an informal, moderated discussion and networking session. Open to all conference participants.

Key takeaways from Design and Engineering perspectives · Building bridges: What did we learn from each other? · Open forum for final questions and reflections · Networking opportunity