Unleashing AI in Your Web Apps
summary
Claims that running small models in-browser eliminates the typical 20ms latency problem, improves privacy and supports offline use by running Gemini Nano via the Prompt API and new Task APIs. Demos use Chrome feature flags, a language detector plus translator/summarizer APIs, streaming outputs, and tools that require the app to execute returned tool calls. A Solo Leveling–style family quest app shows JSON-schema structured output, local session storage, and practical limits: simple tasks stay local, heavy compute falls back to cloud.
AI-written summary · about
chapters (10)
- 0:00🚀 Build Stuff Conference – Get Your Tickets!
- 0:30Introduction: The 20ms Latency Problem
- 3:40Why AI Today Relies on the Cloud
- 6:20Challenges: Latency, Privacy & Offline Limitations
- 9:10In-Browser AI: Running Models Locally
- 12:30New Browser APIs for AI Tasks (Translate, Summarize, Detect)
- 16:00Prompt API & Using Gemini Nano in the Browser
- 20:10Tools & Structured Outputs with Local AI
- 23:00Demo: AI-Powered Web App (Solo Leveling Task System)
- 25:00Future Vision: Hybrid AI Apps (Local + Cloud)
related talks
21mUnlocking Client-Side AI in Your Web App
24mAI right in the browser with Chrome's built-in AI APIs
12mBrowsing, building, and web in the AI era
15mThe future of AI is now: Real-life case studies for on client-side AI adoption in web apps
22mClient side Web AI Agents: Building smarter user experiences for future agentic ...
22mBoring but Effective: WebAI That Ships
16mLessons learned from being customer zero of Chrome's built-in APIs
31mHow Web Apps Are Evolving into AI Agents