Client side Web AI Agents: Building smarter user experiences for future agentic ...
summary
Prototype runs entirely client side using 2 billion parameter Gemma via the Mediap web LLM library, executing on a five‑generation GPU (Nvidia 1070 GPU) with no cloud calls. Built demos include an agent that books flights by calling existing site functions as tools, emits structured JSON for orchestration and follow‑ups, caches large models with a proxy library, extracts video frames offline, and a one‑day web AI DJ using local Cororo TTS and Spotify API calls. Agent memory, tool calling, and local inference all work end‑to‑end.
AI-written summary · about
chapters (11)
- 0:00Introduction to client-side AI agents
- 1:25What is an agentic system
- 3:13Tools and orchestration in agents
- 6:04Demo of agentic flight booking
- 8:57Demo of dynamic flight changes
- 10:21Caching proxy for large models
- 12:28Agent configuration and setup
- 15:19Web AI DJ demo
- 17:03Web AI vs cloud AI
- 18:47Future of agentic web
- 20:31Call to action and resources
related talks
29mClient side Web AI Agents Building smarter user experiences for a future age
24mState of client side AI
02mDiscover Web AI: Client side Agents, Gen AI, and machine learning in the browser
21mUnlocking Client-Side AI in Your Web App
15mThe future of AI is now: Real-life case studies for on client-side AI adoption in web apps
31mBringing and Running AI Agents in the Browser
31mHow Web Apps Are Evolving into AI Agents
33mLocal Models: How to Run AI in the Browser