AI right in the browser with Chrome's built-in AI APIs
summary
Demo shows Chrome running Gemini Nano locally with no network requests, using built-in AI APIs for language detection, streaming translation, and summarization that finish in milliseconds. The page implements the Prompt API with expected-inputs/outputs, JSON-schema response constraints and multimodal image inputs to classify “likes hats” or detect passports, using create/destroy model instances and token-by-token streaming. Live examples and console snippets prove the APIs work end-to-end in-browser without server-side keys.
AI-written summary · about
chapters (5)
- 0:00Chrome’s built-in AI APIs intro
- 1:11Live demo of AI browser features
- 8:44How AI APIs work
- 17:18Prompt API and capabilities
- 23:00Real-world use cases
related talks
21mUnlocking Client-Side AI in Your Web App
15mOverview of Chrome built-in AI
12mBrowsing, building, and web in the AI era
30mUnleashing AI in Your Web Apps
15mThe future of AI is now: Real-life case studies for on client-side AI adoption in web apps
37mFrom zero to built-in AI hero
32mConversations Without Prompts: AI in the Web
53mAI Didn’t Kill the Web, It Moved in!