On-Device and hybrid AI

Building AI agents no longer requires a constant cloud connection or massive server costs. Today, the most innovative models run where your users are — directly on-device.

Join this hands-on workshop to explore Edge AI, an emerging architectural approach that delivers zero per-request costs, full data privacy, and lightning-fast offline performance.

We’ll go beyond simple chat interfaces and dive into building fully functional local-first AI agents using lightweight models like Gemma, Llama, and DeepSeek.

What we’ll cover:

• On-Device Function Calling:
  Enable your agent to interact with local APIs (contacts, calendars, sensors) with zero latency.
• Local RAG:
  Query private documents and notes to generate context-aware responses — without sending data to the cloud.
• Hybrid Strategy:
  Learn a “best of both worlds” architecture that balances edge privacy with cloud capabilities.

The Project:

We’ll build a working prototype using Flutter, Gemma, and GenKit. (Native Android and iOS tracks are also supported.)

By the end of the workshop, you’ll clearly understand the real trade-offs between model size, performance, and capability.