The history of AI agents is traced from simple text completion models to complex, multi-step reasoning systems using frameworks like ReAct.
An AI agent is a system where an LLM decides the application's control flow using reasoning, memory, and tools.
Several composable patterns like reflection, tool use, planning, and multi-agent collaboration provide structured approaches for building agentic systems.
Agentic design patterns are essential building blocks that provide structure, enable composition, and improve the testability of complex AI applications.
Agent evaluation shifts focus from single-attempt model capability to multi-attempt reliability, requiring custom benchmarks to measure real-world task success.
Context engineering expands on prompt engineering by systematically providing the right information, tools, and memory to the LLM at the right time.
You can build agents using Google Gemini through AI Studio, the native SDK, or integrations with open-source libraries like LangChain and CrewAI.
Future agentic systems are predicted to enable zero-cost software creation, personalized multimodal interfaces, and accelerated adoption in robotics.
29:03Jörg Neumann
39:11Thomas Dohmke
26:35Ricardo
27:12Ryan J. Salva
30:01Mario Fusco
25:32Viktoria Semaan
27:07Eric Jadi & Rene Pajta
23:39Dennis Zielke & Rene Pajta
Jobs that call for the skills explored in this talk.
Commerz Direktservice GmbH
Duisburg, Germany
Deutsches Medizinrechenzentrum GmbH
4Com GmbH & Co. KG
Gemma Analytics