AI Agents for Architects

Planning, Coordinating, and Executing Tasks with Agentic AI Systems

Agentic AI systems plan tasks, execute steps, and coordinate tools. This training combines solid foundational knowledge with practical transfer impulses and demonstrates how agentic AI systems can be architecturally conceptualized and positioned. The focus is on the core building blocks of agentic systems and the key questions that should be addressed early in design and operation.

Upon completing the training, participants will be able to evaluate agentic AI initiatives in a structured way and justify architecture decisions confidentlyfrom initial assessment to preparation for productive environments.This course is aimed at anyone who wants to efficiently integrate AI solutions into existing IT landscapes while focusing on scalability, security, and maintainability.

What can you expect?In the training AI Agents for Architects, you will gain practical architectural knowledge to understand agentic AI systems and position them effectively in real projects. You will learn how agents are designed as a combination of central building blocks and the role of context, memory, and state for reliable behavior. The focus is on interoperability (e.g., using standards like MCP), as well as quality, observability, and requirements for security and scalability in productive IT environments.

Participant prerequisites

• Solid understanding of software architecture and the design of software systems, APIs, and integration interfaces

• Basic understanding of GenAI and LLM usage, e.g., context windows, tokens, and RAG as a concept

Your benefits

• Understand the agentic paradigm and clearly distinguish agentic AI systems from classical GenAI scenarios

• Learn how agents are architecturally structured and how core building blocks fit into a robust overall system

• Acquire practical decision logic for context engineering, memory, and state management

• Learn how tool integration and interoperability are structured in agent architectures, including MCP and multi-agent systems

• Receive criteria and approaches to evaluate the quality, robustness, and security of agentic systems

• Learn how to prepare agentic systems for productive environments with a focus on observability, scalability, and secure operation

Detailed training content

From Model to Agent System

        Understanding what distinguishes agentic AI systems and how this changes architectural decisions

        Overview of the core building blocks and their interplay in agentic systems

Managing Context, Memory, and State

        The role of context, memory, and state for reliable behavior and common pitfalls

        Approaches for handling context so that systems remain stable under real-world conditions

Tool Integration and Collaboration in Agent Landscapes

        How agents use tools and why interoperability should be considered early

        MCP and multi-agent systems as guidance for standardized collaboration and extensibility

Making Quality Visible

        How to meaningfully think about and evaluate quality in non-deterministic systems

        Observability as the foundation for traceable behavior in operation

Safe Deployment into Production

        What counts for production readinessfrom scalability to securitywithout losing oversight

        Guidelines for operations that can be integrated into existing development and release processes