Multi Agent Architecture using ADK
summary
A clonable ADK codebase demonstrates a working multi‑agent agent-to-agent protocol that wires a judge agent and SQL agent with a masking layer, observability and Cloud Run deployment; the workshop shows ADK invoking tools (internet, DB, front end) to plan, act, observe and loop, claims ~80% of common enterprise tasks are automatable, cites forecasting as a trillion‑dollar pain, and surfaces real failures (an auto‑code push that wiped a repo) to stress evaluation.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:20Multi-agent ecosystem
- 12:50Companies building AI agents
- 20:16Automation agents examples
- 28:05Reasoning agents workflows
- 31:58Agent infrastructure benefits
- 41:24Agent complexity risks
- 50:16Reasoning AI agent anatomy
- 57:09ADK architecture
- 1:04:58ADK live demo
- 1:08:11Q&A
related talks
54mCreate Multi-Agent Systems with ADK
01h 03mSupercharge your AI agents: Inside the new ADK integrations ecosystem
45mUnifying AI experience for customers: Building multi-agent systems with Google ADK
24mAI Agent Swarms Are Amazing
28mThe Put That Sh*t On Agent: Building Multi-Agent AI Systems That Actually Talk to Each Other*
26mAI powered automation & multi-agent orchestration
57mCoding with ADK to beat a text adventure game
02h 59mStandards and AI Agents: an Overview of MCP, A2A, ADK and More