May 19
Loading video player...
GitHub repo → https://goo.gle/3Pn0Z01 [Codelab] Building ADK Agents with Skills and Tools → https://goo.gle/4wB4515 Explore a high-scale agentic AI-powered simulation sandbox → https://goo.gle/4nEraMv Discover how to build enterprise ready AI agents using the open source Agent Development Kit (ADK 2.0) and Gemini models, as showcased in the Google Cloud Next '26 Dev Keynote. Learn to efficiently manage context by leveraging modular "skills" and remote MCP servers—including Google Maps and Workspace—to orchestrate complex, real world tasks like generating mathematically perfect GeoJSON routes. Get hands on with the open source "race condition" GitHub repository and codelabs to deploy your own scalable, multi-agent workflows on Agent Runtime, Cloud Run, or GKE today. Chapters: 0:00 - Intro 0:53 - What is Agent Development Kit (ADK)? 2:07 - Equipping agents with skills and tools 3:54 - [Demo] The Las Vegas marathon simulation 4:57 - Calculating routes with GIS code 5:53 - Integrating the Google Maps MCP server 7:00 - Converting documents into non-deterministic criteria 8:32 - Exploring the open source GitHub repository 10:13 - Code walkthrough 12:34 - Mapping & grounding skills 13:58 - Race director skill 14:37 - [Codelab] Building ADK Agents with Skills and Tools 16:07 - Summary and getting started Watch more Google Cloud Next 2026 → https://goo.gle/next-talks-2026 🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech #GoogleCloudNext Speaker: Mofi Rahman Products Mentioned: Google Cloud Next, Google Kubernetes Engine, Agent Development Kit, Google Maps, Google Workspace, Cloud Run, Gemini