What was your expectation? That your prompt would trigger a web search, first, before the introspection of past conversations and a training set recall?

How did Sonnet 4.6 respond that was objectively better for your use case?