We successfully conducted a hands-on workshop on building AI Agents using Google’s Agent Development Kit (ADK), designed especially for developers and Google Hackathon participants.

In this 60-minute interactive session, participants learned how to go beyond traditional chatbots and build agentic systems that can reason, call tools, and act intelligently using Python. The workshop focused on a code-first, practical approach, enabling attendees to build and run AI agents locally on their machines.

By the end of the session, participants had a working AI agent capable of handling user queries, invoking tools, and making intelligent decisions—perfect for hackathon-ready projects.

What Participants Learned

• What AI Agents are and how they differ from traditional ML models and chatbots

• Overview of Google Agent Development Kit (ADK) and its architecture

• How an agent “thinks”: prompts, instructions, reasoning, tool calls, and final output

• Setting up ADK locally and creating an agent project

• Writing a root agent with clear instructions and behavior policies

• Running and testing agents via CLI and ADK Web UI

• Enhancing agents with custom tools (functions & APIs)

• Building multi-tool agents for real-world use cases and hackathons

Key Takeaways

• Agents are systems that reason and act, not just chat interfaces

• ADK provides a production-ready framework for building intelligent agents

• Tool integration is the key to building powerful, hackathon-winning solutions

• Participants left with a working AI agent and a clear roadmap for scaling it