Build Your Own AI Sidekick with Flutter
summary
Hands-on build of a tiny Flutter AI sidekick using Firebase AI (Gemini developer API) and optional Vertex AI, with boilerplate code and Chrome/web debugging recommended; attendees set up a new Firebase project on the free tier, add the Gemini SDK, and connect generative models (Flash, Gemini 2.5 variants) to a simple chat UI. Workshop shows multimodal image inputs, enforces structured JSON output for widget-ready responses, demonstrates streaming limitations with JSON and the power of function calling to let the model invoke app logic; many quick demos (20 Guesses, game creator — winner, pet analyzer, image transformer, recipe app) work live within the session time.
AI-written summary · about
chapters (22)
- 0:00Workshop introduction
- 3:53Flutter project setup
- 12:34Basic UI for AI sidekick
- 22:02Gemini model config
- 26:57Multimodal inputs
- 29:27JSON output enforcement
- 34:00Streaming responses
- 36:14Function calling
- 38:01AI sidekick ideas
- 51:3420 guesses game
- 53:06Game creator app
- 54:53Public transport buddy
- 57:10Pet analyzer tool
- 58:13Petting zoo builder
- 1:00:20Image transformation app
- 1:01:26Emoji story spinner
- 1:02:11Melody tracker
- 1:03:19Recipe generator
- 1:04:04Risk game assistant
- 1:07:21Sarcastic friend bot
- 1:08:25Audience voting
- 1:10:53Closing remarks
related talks
36mEmpowering Your Flutter Apps with AI
38mEnhance the customer experience
42mDart and Flutter and AI, Oh My!
14mPrompt to production: Building an AI app with Flutter
01h 19mFull Stack AI Features
42mMore Flutter and Dart and AI, Oh My!
47mAgentive Experiences With Flutter: What, How, and Why Bother?
10mMeet the Speakers