Claude is likely talking to sub-agents, although I have yet to see the questions written out. But sometimes the agents do well, in which case you will see "Excellent research!" or similar printed to the conversation history. It's my guess these conversations that Claude has with various specialists agents will be hidden from users soon, but it's an interesting look behind the scenes, indicating that we don't manually need to set up specialist agents and orchestrate this as users, Anthropic is likely already doing it automatically.