Google DeepMind's native multimodal agents demoed

Patrick Löber shows how drawing arrows on a map prompts Gemini to generate an image of the Golden Gate Bridge, leveraging world understanding. The talk explores any-to-any multimodal capabilities in agents.