Build a powerful AI agent with Google's Agent Development Kit (ADK) and the Box Content Cloud
2Get started and build your own agent here → https://github.com/google/agent-development-kit
In this step-by-step tutorial, we show you how to create a smart agent that connects to Box, searches files, reads documents, and uses Box AI to answer questions about your content.
Whether you are a developer looking to build agentic systems or an architect designing content-aware AI, this video provides a technical walkthrough of the entire process. We cover everything from creating foundational tools to assembling a complete agent hierarchy with Google's ADK and the Box AI toolkit.
Key takeaways:
↳ Build focused, single-purpose tools that connect your agent to the Box API
↳ Use the Box AI Agents Toolkit to handle authentication and text extraction
↳ Assemble a multi-layered agent using a sub-agent for logic and a root agent for orchestration
↳ Understand the difference between the custom `BaseAgent` approach and the simpler `LLMAgent` definition
↳ Run and interact with your agent through the ADK web interface
Chapters:
00:00 - Introduction
00:54 - Google's Agent Development Kit (ADK) Overview
02:24 - Building the Foundational Tools for Box
06:41 - Assembling the Sub-Agent and Root Agent
10:27 - A Simpler Approach: The Direct Root Agent
14:40 - Demo: Solving an Accounts Payable Challenge
17:10 - Key Development Takeaways
Key Links:
Google ADK Docs: https://google.github.io/adk-docs/
Google Agent Development Kit (ADK): https://github.com/google/adk-python
Code repo for video: https://github.com/box-community/google-adk-box-agent
Box announcement: https://blog.box.com/content-action-box-and-google-cloud-introduce-ai-agents-enterprise