AI Chat
AI Chat builds on AI Describer however permits customers to ask questions on their present view, previous views, and close by geography. The chat agent makes use of Google’s Multimodal Reside API, which helps real-time interplay, perform calling, and briefly retains reminiscence of all interactions inside a single session. We monitor and ship every pan or motion interplay together with the person’s present view and geographic context (e.g., close by locations, present heading).
What makes AI Chat so highly effective is its potential to carry a brief “reminiscence” of the person’s session — the context window is ready to a most of 1,048,576 enter tokens, which is roughly equal to over 4k enter photographs. As a result of AI Chat receives the person’s view and placement with each digital step, it collects details about the person’s location and context. A person can nearly stroll previous a bus cease, flip a nook, after which ask, “Wait, the place was that bus cease?” The agent can recall its earlier context, analyze the present geographic enter, and reply, “The bus cease is behind you, roughly 12 meters away.”