Building Agentic Apps with Google ADK - Part 1 - Basic Agentic Apps

An Agent is a self-contained execution unit designed to act autonomously to achieve specific goals. Agents can perform tasks, interact with users, utilize external tools, and coordinate with other agents. Key features that defines an agentic AI application are memory, session and state management, tool use, planning and multi-step reasoning among others. Using Google ADK, we will explore some of these basic agentic constructs and build out a simple agent that can interface with the external world in order to perform a provided task.

Key Takeaways:

• Understand what makes an agent or an agentic application
• Learn the features of Google ADK and how to use it to develop a basic agentic application
• Learn how you can enhance model capabilities with tools, and see how the model reasons to leverage tool calling capabilities

Prerequisites:

A laptop with internet connection and a standard browser

Access to Google Colab (https://colab.new/)

A Gemini API key (https://aistudio.google.com/apikey)

With two decades of experience in the technology sector, Prashanth Subrahmanyam heads Developer Advocacy for APAC at Google Cloud. For over ten years, he has focused on cloud and web technologies, with a significant emphasis on AI/ML for the past five years.

Prashanth is a passionate advocate for the use of AI to increase developer productivity, build impactful products, and modernize existing applications. His expertise spans Big Data, Data Engineering, Data Analytics, AI/ML, DevOps, CI/CD, Serverless, and API platforms. In his free time he enjoys gardening and spending time with the plants, playing video games, and wonders how the fish in his aquarium spend their whole day without mobiles and social media.