Build agents with Gemini API
summary
Gemini 3.5 Flash drives a workflow that builds and runs agents end‑to‑end: AI Studio generates a native Kotlin app or a Node.js full‑stack project, exports it to a local IDE, and installs to Android in one click; the demo shows a working daily stretch app and an AI Talk Radio sequence that produces TTS, music and cover art. The Interactions API is GA and supports mixing Google Search with custom functions, async background runs and persistent remote sandboxes (~8–10s startup) behind a network proxy for credential safety. Gemini 3.5 Flash AI Studio full‑stack Node.js export interactions API GA
AI-written summary · about
chapters (9)
- 0:00Gemini platform overview
- 3:06AI studio features
- 7:46Enhanced UI and UX tools
- 11:03Getting started with AI studio
- 12:42Building and deploying apps
- 17:03Interactions API introduction
- 23:03Remote agents and environments
- 26:35Gemini API CLI and agent demos
- 29:22Finalizing agents and wrap up
related talks
39mBuild the new generation of AI Agents
15mBuilding with the Gemini API and AI Studio
01h 45mAI Engineering with the Google Gemini 2.5 Model Family
40mDevelop faster with Gemini Agents in Android…
18mIntroduction to Gemini APIs and AI Studio
01h 48mBuilding Conversational Agents
18mLet's test Gemini, Google's AI in an Android application!
01h 55mPrompt to Pipeline: Building with Google's Gen Media Stack