Event Overview
Date: May 2025
Location: Google Office, Bengaluru
Today, I attended the "Build with AI: Agent Development Kit (ADK) Hands-on Workshop" organized by Google and Deep Tech Stars at Google Office, Bengaluru —and what a day it was! Despite falling on a Sunday during a long weekend, the Google office buzzed with nearly 200 developers, all united by curiosity about Google’s newly launched ADK and the future of AI agents.
The event started with Nihal Kashinath sharing Deep Tech Stars’ vision for fostering innovation, followed by Nikhil Rana demonstrating Google AI Studio’s Stream Realtime feature. Watching an AI model generate responses in real-time felt like glimpsing the future. Prasad Seth then took the stage to talk about GDG communities and how developers leveraging tools like ADK could earn spots at major Google events—a motivating nudge for anyone building with AI!
Hands-on Workshop
The hands-on workshop, led by Nikhil Rana, was the highlight. We rolled up our sleeves to build a weather_time agent from scratch: setting up projects, installing libraries, enabling Vertex API, and using Cloud Shell and Google Shell Code Editor to bring the agent to life. What stood out to me was ADK’s user-friendly web interface and its built-in debugging tools. The tracing functionality and Agentic Eval features (similar to LangGraph) felt like game-changers for monitoring agent behavior—no more guesswork in troubleshooting!
Model Context Protocol (MCP)
Later, Bhavesh Wadhwani gave an insightful deep-dive into Model Context Protocol (MCP). He explained how MCP tackles the complexity of managing countless tools for LLMs, using a practical AirBnB demo to show its real-world value. I left that session bookmarking the MCP server resources he shared and curious about integrating it with ADK via the A2A protocol.
Community & Takeaways
What I loved most was how approachable everything felt. Whether it was Nihal breaking down ADK’s evaluation methods or Nikhil Rana patiently guiding us through code hiccups, the focus was on learning by doing. And while the tech was impressive, the real takeaway was the community—seeing so many developers trade weekend plans for AI exploration was inspiring.
A big thank you to Nihal Kashinath, Prasad Seth, Nikhil Rana, and Bhavesh Wadhwani for organizing such a thoughtful event. If you’re experimenting with ADK or MCP, I’d highly recommend checking out the resources they shared—it’s rare to find guides that balance depth with clarity so well.
Here’s to building smarter agents (and maybe seeing some of you at another Google event soon)! 🚀