Author's): Laura Verghote
Originally published in Towards Artificial Intelligence.
Create reasoning agents with the Agent Development Kit (ADK), coordinate Think-Act-Observe loops, and discover real-world multi-agent design patterns with Gemini.
I'm participating this week 5-day intensive Google course for AI agents. This is a week-long, hands-on sprint aimed at building real AI agents. Created by Google ML researchers and engineers, the program combines theory, programming exercises, and live discussions. Its goal is to help developers move beyond LLM demos and build production-ready systems that integrate models, tools, orchestration, storage, and evaluation.
The first day of the Google AI Agents Intensive introduces the basic concepts of creating AI agents, focusing on the orchestration layer that governs agent decision-making and tool use. The course emphasizes the importance of context engineering in improving agent performance, and hands-on exercises are provided through Kaggle programming exercises. Topics such as the classification of agents based on their capabilities and different patterns for multi-agent systems are also discussed, with a clear shift from a single program to a collaborative framework that dynamically refines processes.
Read the entire blog for free on Medium.
Published via Towards AI
