Privacy-first in-browser Generative AI web apps
summary
Live demo runs fully offline on a Surface Laptop with an NPU, hosted on Vercel, and shows image classification working with a visible CPU/NPU spike; model loads cache to the browser (~50–100 MB example) so features persist without network. Explains the ecosystem chain: WebNN as the only browser API that can access NPUs today, higher layers like ONNX Runtime Web and transformers.js, and the practical caveats — canary browsers, flags, driver updates and model sizes ranging from dozens to hundreds of megabytes.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 2:51In-browser generative AI matters
- 6:22Local AI workload options
- 8:51Web Neural Network API standard
- 13:08WebNN and frameworks
- 15:16Hardware for local AI
- 18:09Privacy and performance benefits
- 20:20Developer advantages
- 22:30WebNN prerequisites
- 23:36High-level AI frameworks
- 26:49Transformers.js and ONNX Runtime Web
- 29:18Summary and call to action
related talks
45mPrivacy-First In-Browser Generative AI Web Apps
46mThe Ultimate Guide to Browser-Based AI: Privacy-First, Offline-Ready and Future-Proof
41mEscaping the Cloud: High-Performance AI in your Browser
21mUnlocking Client-Side AI in Your Web App
16mThe Web is AI Ready—maximize your AI web development with WebNN
18mWeb AI on next generation AI PCs
17mRunning Intelligence at the Edge: Web AI in Action
31mGenerative AI Directly in My Browser, Is It Possible? (spoiler, yes 😉)