Uploaded June 2026 | Updated September 2026, 2 weeks ago
Google I/O '26 agentic releases β https://goo.gle/4v6NdOP
A new era for AI Search β https://goo.gle/4xl0hkS
Ready to move past basic AI boilerplate and see agents handle a dependent chain of complex engineering problems? This video showcases long-running AI agents executing multi-day pipelines to build non-trivial software, from an active window managed desktop OS to a fully optimized 3D web application. Join the author of Beyond Vibe Coding, Addy Osmani, as he discusses the AI agents behind the magic.
Watch more The Agent Factory: Developer Power-Ups β https://goo.gle/Supercharge-Devs
π Subscribe to Google Cloud Tech β https://goo.gle/GoogleCloudTech
#DeveloperPowerUps
Speakers: Addy Osmani
Products Mentioned: Agent Development Kit
Google I/O '26 agentic releases β https://goo.gle/4v6NdOP
A new era for AI Search β https://goo.gle/4xl0hkS
Ready to move past basic AI boilerplate and see agents handle a dependent chain of complex engineering problems? This video showcases long-running AI agents executing multi-day pipelines to build non-trivial software, from an active window managed desktop OS to a fully optimized 3D web application. Join the author of Beyond Vibe Coding, Addy Osmani, as he discusses the AI agents behind the magic.
Watch more The Agent Factory: Developer Power-Ups β https://goo.gle/Supercharge-Devs
π Subscribe to Google Cloud Tech β https://goo.gle/GoogleCloudTech
#DeveloperPowerUps
Speakers: Addy Osmani
Products Mentioned: Agent Development Kit










![How to design a multi-agent system that skips the LLM
Github repo β https://goo.gle/race-condition
Previous episode β https://goo.gle/marathonagent
A thousand AI agents run a marathon, and almost none of them ever call the LLM.
In this multi-agent system deep dive, Casey West breaks down the one architectural decision behind Race Condition: a 1000-agent system built on Googles Agent Development Kit (ADK).
The question every AI engineer is wrestling with: when do you let an LLM decide, and when do you just write the code in a multiagent system? We trace one decision end to end, planning a marathon route, then show how the same idea (skip the LLM where you dont need it) scales to a thousand agents running on deterministic code.
What youll learn:
* When to use an LLM vs deterministic logic
* The before_model_callback trick, keep the agent, skip the model
* Why route planning is deterministic (NP-hard + the Spine & Sprout algorithm)
* How 1,000 autopilot runners make 0 LLM calls
* Where the tokens actually go (the AI decides, the code runs)
* Scaling 1,000 stateless sessions with Redis
Chapters
00:00 - Intro: 1,000 AI agents that dont call the LLM
00:41 - When should an agent use an LLM?
01:02 - [Demo] Planning a marathon route
01:59 - Why Google Maps cant route a marathon
05:08 - Why the LLM Is the wrong tool (NP-hard)
05:40 - The deterministic spine & sprout algorithm
06:58 - Using AI Studio to choose the algorithm
09:00 - The trick: Skip the LLM with a callback
12:26 - before_model_callback β the reveal
17:50 - Autopilot runners: 1,000 agents, 0 LLM calls
21:31 - How many tokens? Where they actually go
23:28 - The second cost: Session state & redis
29:05 - Wrap up
More resources:
Google Agent Development Kit (ADK) β https://goo.gle/3PItVzL
Google ADK Community (Redis session service) β https://goo.gle/4ugzmUw
Agent Runtime β https://goo.gle/4nXDhnX
Google Cloud Memory Store β https://goo.gle/4nXxBtT
Agent2Agent Protocol (A2A) protocol β https://goo.gle/4u5x8HF
Casey West on LinkedIn β https://goo.gle/4dXnsJr
Annie Wang on LinkedIn β https://goo.gle/43GCXAo
Watch more Hands on AI β https://www.youtube.com/playlist?list=PLIivdWyY5sqKnJOvP89yF8t9mWuzMTcbM
π Subscribe to Google Cloud Tech β https://goo.gle/GoogleCloudTech
#AIAgent #GoogleADK #Gemini #MultiAgentSystem #AgenticAI #GoogleCloud
Speakers: Casey West, Annie Wang
Products Mentioned: Google Agent Development Kit, Gemini API, Agent Runtime, Google Cloud Pub/Sub, AlloyDB, Agent2Agent Protocol How to design a multi-agent system that skips the LLM](https://i.ytimg.com/vi/Fzd0BWMH65s/mqdefault.jpg)