Google Antigravity Demo: Vibecode An Ambient AI Agent with ADK & Agent CLI (Full Code Lab)
Full hands-on demo of Google Antigravity, Google's new agentic development IDE. I share my screen and build a corporate expense agent end to end — an ambient, event-driven agent that takes employee expenses and, based on whether they're over or under $100, auto-approves them with deterministic code or routes them through LLM security screening. This follows the "VIP Code and ADK 2.0 Ambient Agents" code lab from Google's 5-Day AI Agent Intensive course. Everything is built with Google's ADK (Agent Development Kit) and scaffolded with Agent CLI, using a Google AI Studio key for model access. You'll see the real experience — including hitting free-tier quota limits and switching models (yes, Antigravity runs Claude too) to keep building. 🔧 What's in the video: • What Antigravity is and how it compares to other agentic IDEs • Installing ADK skills and the Agent CLI toolchain • Scaffolding an ADK 2.0 project from a natural-language prompt • Building a stateful graph workflow: extract → route → auto-approve/reject • Adding PII scrubbing + prompt-injection defense • Testing the agent in the ADK playground ⏱️ Timestamps 0:00 – Intro: what we're building 0:43 – The code lab & use case (ambient expense agent) 3:01 – Game plan + prerequisites (Python 3.11, uv, AI Studio key) 3:58 – Configuring Antigravity & loading ADK skills 7:13 – Skills installed — setting up the workspace 9:55 – Initializing the ADK project with Agent CLI 12:33 – Prompting with the use case & scaffolding the project 14:00 – Touring the generated code (agent.py, env config) 15:41 – Credentials + ADK cheat sheet 19:53 – Implementation plan: the stateful graph core 21:21 – Core logic: extract expense, routing, auto-approve/reject 25:20 – Out of tokens! Switching models to Claude 25:54 – Adding security: PII scrubbing & prompt-injection defense 28:14 – Reviewing the security checkpoint code 30:47 – Testing in the ADK playground 33:02 – Hitting quota again & final thoughts 🔗 Resources • Google Antigravity: https://antigravity.google • Google AI Studio: https://aistudio.google.com • ADK (Agent Development Kit): https://google.github.io/adk-docs • 5-Day AI Agent Intensive code lab: https://codelabs.developers.google.co... • My skills explainer video: • Google AI Intensive Day 3: Giving AI Agent... 💬 Comment if you want a follow-up when quota resets — part 2 finishes the ambient trigger + LLM-as-judge eval. Subscribe for more hands-on AI agent builds. #GoogleAntigravity #ADK #AIAgents #AgentDevelopmentKit #GoogleAI #AIStudio #Gemini #AICoding #VibeCoding

Which Antigravity surface should I use? 2.0 vs CLI vs IDE vs SDK
![How I Use AI to Automate Google Ads [Tech skills NOT required]](https://i.ytimg.com/vi/5q5YHtLFpx8/hqdefault_custom_1.jpg?sqp=CNyV0NIG-oaymwEjCNACELwBSFryq4qpAxUIARUAAAAAGAElAADIQj0AgKJDeAE=&rs=AOn4CLBEmt98U6BDEULCrwETU4nhxrV4Nw)
How I Use AI to Automate Google Ads [Tech skills NOT required]

The Best Local Agentic Coding Workflow (Complete Guide)

Why Google Just Gave Away Gemma 4 for Free

Building The Voice Engine

Guide to Agentic AI – Build a Python Coding Agent with Gemini

Building an AI Dark Factory: A Codebase That Writes Its Own Code, Live

China Is About To Pop The AI Bubble

Full Archon Guide - Build AI Coding Harnesses That Actually Ship (LIVE)

No Boss, No Money: The Raw Reality of China’s Gen-Z Freelancers

End of Gemini CLI - Welcome to Antigravity 2.0

Learn the basics of Google Antigravity

Model Context Protocol (MCP) Explained for Beginners: AI Flight Booking Demo!

MIT Just Revealed the AI Bubble's Fatal Flaw

Claude Code Tutorial - Build Apps 10x Faster with AI

Creator of C++: Bell Labs, Negative Overhead Abstraction, Mistakes | Bjarne Stroustrup

What rebuilding AlphaGo teaches us about self-play, RL, and future of LLMs - Eric Jang

CLAUDE CODE ADVANCED FULL COURSE (3 HOURS)

Andrej Karpathy: From Vibe Coding to Agentic Engineering w/ Stephanie Zhan

