Your chance to start building AI agents from the absolute basics

Google Cloud has introduced Agent Valley, a free five-week live learning program aimed at beginners interested in building AI agents without requiring prior machine learning or software development experience. The course uses a low-poly virtual world as a practical workspace, allowing participants to create and interact with their own agent systems in real time. Sessions are led by Annie Wang, a Google Developer Relations engineer, who designed the curriculum to be fully hands-on and accessible to newcomers. The environment is provided by Google, eliminating setup barriers and enabling immediate participation.

Week 1 focuses on summoning a custom low-poly companion and maintaining consistent memory and traits during interactions, introducing foundational control concepts. Participants describe their companion and observe its behavior in a split-screen workspace alongside a live Runtime Inspector that displays the AI’s decision-making and resource usage in real time. The course emphasizes practical application over theoretical instruction, with each week building on the previous lessons.

Week 2 shifts to project decomposition, teaching participants how to break complex tasks into smaller components so multiple AI assistants can collaborate without conflicts. This module introduces parallel processing and coordination strategies, using a virtual buildyard as the learning environment. The exercises are designed to mirror real-world scenarios where AI systems must manage interdependent tasks efficiently and reliably.

Week 3 centers on coordination by simulating a virtual shop where participants write code to handle transactions and returns smoothly. The focus is on reliability and error handling, ensuring that agent interactions remain stable under varying conditions. Subsequent weeks progress to memory management, where participants learn to help their agents retain past details without confusion, and real-time responsiveness, culminating in a final project that integrates all learned concepts.