Agentic AI: The New Software Paradigm
This course introduces the emerging paradigm of Agentic AI — systems that move beyond simple automation to exhibit adaptive, autonomous, goal-driven behavior. You’ll learn how these systems are designed and what components (LLMs, memory, context engineering, agentic control flow, tools) make reliable real-world agents possible.
Overview
This course explores how modern AI systems evolve from “text-in/text-out” prompting into full agentic systems that can plan, take actions, interact with their environment, and coordinate workflows. You’ll build a clear mental model of agentic architecture — covering LLM foundations, memory, context engineering, tool use, and multi-agent patterns — then connect these ideas to today’s platforms and frameworks and to the practices that make agent behavior measurable and improvable (evals, observability, fine-tuning, distillation).
Which topics will be covered?
- Agentic AI fundamentals and how it differs from prior AI/automation systems
- Agentic architecture: control flow, decision-making loops, and memory (short-term context + long-term retrieval)
- Context engineering + tools + code execution as the core building blocks for real agent workflows
- Multi-agent systems: roles and coordination patterns; frameworks such as AutoGen/CrewAI and the broader ecosystem
- Deploy, evaluate, and improve agents: local models, evals, observability, fine-tuning and distillation
What will I achieve?
By the end of the course, learners will be able to:
- Explain what Agentic AI is and how it differs from prior AI systems and traditional software approaches.
- Describe the core architecture of agentic systems (context engineering, memory, and agentic control flow/orchestration, etc.).
- Identify capabilities and use cases of autonomous and multi-agent systems and how they enable real workflows.
- Understand the engineering discipline required to make agents reliable: evaluation, observability, and continuous improvement (including fine-tuning/distillation).
- Recognize the broader impact of agentic systems on products, workflows, industries, and the future of AI assistants/orchestration.
Which prerequisites do I need to fulfill?
Basic familiarity with AI/ML or Large Language Models is recommended but not required; the course introduces key concepts in an accessible, structured way for both technical and non-technical learners.
The underlying project was funded by the Federal Ministry of Research, Technology and Space under the funding code “KI-Servicezentrum Berlin-Brandenburg” 16IS22092. Responsibility for the content of this page remains with the author.
