These kinds of multimodal agents are being worked on by Google, OpenAI, and others, and now Perplexity has announced ... or tap the keyboard icon in the lower right corner to type instead.