Developer recap of
summary
Live demo shows a three‑agent system that plans a Las Vegas marathon for 10,000 runners by combining an Agent Development Kit (ADK) planner, a GIS Python skill that calculates GeoJSON race routes, a Google Maps MCP mapping skill, and a text‑to‑policy race‑director skill converted from Google Docs; an evaluator runs real‑time checks using Gemini 3.1 Pro evaluation and also validates A2UI outputs, while 1,000 runner sessions stream over Pub/Sub; all code is open source in the Race Condition repo with codelabs.
AI-written summary · about
chapters (11)
- 0:00Holding slide & countdown
- 2:24Stephanie Wong introduction
- 3:55Mofi Rahman
- 21:01Ivan Nardini x Casey West
- 41:37Craig Labenz x Khanh Nguyen
- 48:06Lucia Subatin x Jack Wotherspoon
- 57:23Cynthia Wang
- 1:08:34Aja Hammerly x Jigyasa Grover
- 1:18:55Rody Davis x Aja Hammerly
- 1:26:20Richard Seroter x Aja Hammerly
- 1:35:51Wrap up
related talks
54mCreate Multi-Agent Systems with ADK
40mAI Agents in Practice
07h 47mAI Agents Conference 2026 - Day 5
45mBuild an agent with Mapbox MCP
16mCreate agent skills for on-device generative AI
45mUnifying AI experience for customers: Building multi-agent systems with Google ADK
25mHow Google DeepMind Runs Agents at Scale
01h 30mHow to build AI Agents: A step by step guide to creating AI agents