Build an AI Agent with Gemini 3

Gemini 3 is our most intelligent model yet that helps you bring any idea to life. In this video, you’ll learn how to build an AI Agent with Gemini 3 Pro and the Agent Development Kit (ADK). You’ll explore how to build a single agent with access to Google Search as its tool, and interact with it using a web UI. We will also learn how Gemini 3 Pro uses Thought Signatures and we will explore its chain of thought when the agent receives a prompt.

Resources:
Getting started with uv → https://goo.gle/4a7M7KG
ADK Google Search Tool docs → https://goo.gle/3LGFKUR
Github Repo for the AI Agent built with Gemini 3 Pro →
https://goo.gle/4pq0pep

Timestamps:
0:00 – Introduction to Gemini 3 Pro
0:32 – Initializing the Project & Installing ADK and GenAI libraries
1:25 – Setting up Google AI Studio’s API Key & Creating Agent Scaffolding with ADK Create
2:00 – Implementing the Google Search Tool
2:33 – Configuring the Gemini 3 Pro Model
3:14 – Deploying the Agent to ADK Web
4:29 – Testing the Agent
5:24 – Thought Signatures and Analyzing Chain of Thought
6:40 – Conclusion

Speaker: Smitha Kolan
Products Mentioned: Gemini, Google AI, Agent Development Kit (ADK)

source