Google’s Agents CLI is here: Build, eval, and deploy AI agents in minutes | The Agent Factory

Google Cloud Tech · Beginner ·🤖 AI Agents & Automation ·1w ago
Agents CLI in Agent Platform → https://goo.gle/3OAExQF ADK.dev → https://goo.gle/4mVq8LI Developer’s guide to building ADK agents with skills → https://goo.gle/41XMB0T Build ADK agents with Google MCPs → https://goo.gle/4e98xxu In this episode of The Agent Factory, we sit down with Shubham Saboo, Senior AI Product Manager at Google and the creator behind the viral "Awesome LLM Apps" repository (105K+ stars), to bridge the gap between AI hype and reality. You’ll learn why the most successful agents aren't built on complex code, but on clear communication and simple architectures. Watch as Shubham demos the brand new Agents CLI in Agent Platform and showcase everything new in ADK. Learn how to scaffold, test, and deploy an agent directly from your terminal using simple English prompts. From building a "caveman" text compressor to a sophisticated multi-agent system that roasts GitHub Pull Requests, you’ll see firsthand how Google’s new agent platform handles the heavy lifting of environment config, YAML files, and cloud deployment. By the end of this video, you’ll walk away with a blueprint for building "ambient agents" that run 24/7, handle drop connections with "resumable" flags, and interact with the world via Google Search tools. Whether you're a hardcore developer or a technical founder, you'll see how to treat AI agents like interns to get the best possible output. Chapters: 0:00 - Introduction: Meet Shubham Saboo 1:54 - The story behind 105,000 GitHub stars 4:52 - How AI agents evolved from GPT-3 to now 8:14 - What is Google’s New Agent CLI? 10:06 - Demo: Building the "Caveman Compressor" agent 17:45 - How to run AI evaluations (Evals) automatically 19:52 - Adding Google Search tools to your agent 23:20 - Demo: Building a Multi-Agent PR Roaster 26:45 - Live Roast: AI Agent roasts my code 30:14 - New in ADK: Graph-based workflows vs. Prompts 33:28 - Resumable Agents: Surviving production reality 34:40 - Ambient Agents: Building agents that run 24/7 36:32 -
Watch on YouTube ↗ (saves to browser)
Sign in to unlock AI tutor explanation · ⚡30

Related AI Lessons

Beyond the Prompt: Building Your First Production-Ready Agent with p-agent and MCP
Learn to build a production-ready agent using p-agent and MCP, enabling autonomous systems that interact with the real world
Medium · Python
Can AI Agents Mainstream the Security Data Lake?
AI agents may revolutionize security data lakes, making them more mainstream and efficient, which is crucial for cybersecurity teams
Medium · Cybersecurity
Your Phone’s Secret AI Sensors
Discover the secret AI sensors in your phone and how they collect data without consent, sparking a growing privacy debate
Dev.to AI
How 10 Agencies Are Running Claude for Their Clients Without the Billing Nightmare
Learn how 10 agencies are managing Claude for their clients without billing issues, using strategies like automated workflows and cost-splitting methods
Dev.to AI

Chapters (12)

Introduction: Meet Shubham Saboo
1:54 The story behind 105,000 GitHub stars
4:52 How AI agents evolved from GPT-3 to now
8:14 What is Google’s New Agent CLI?
10:06 Demo: Building the "Caveman Compressor" agent
17:45 How to run AI evaluations (Evals) automatically
19:52 Adding Google Search tools to your agent
23:20 Demo: Building a Multi-Agent PR Roaster
26:45 Live Roast: AI Agent roasts my code
30:14 New in ADK: Graph-based workflows vs. Prompts
33:28 Resumable Agents: Surviving production reality
34:40 Ambient Agents: Building agents that run 24/7
Up next
This AI agent writes VIRAL posts
Diandra Escobar
Watch →