Intent to Prototype: Embedding API
Summary
The Embedding API is a proposed Web Platform API for on-device generation of high-dimensional embeddings, enabling semantic search, on-device RAG, and real-time content intelligence offline. It aims to reduce latency, cost, and privacy trade-offs by using a shared on-device model across origins, avoiding cloud dependencies. The proposal discusses use cases, interoperability discussions, and the Chrome platform status context for on-device AI capabilities.