“Azeem Azhar assesses the quality of reports from his 'deep research' AI agent as comparable to the output of two junior analysts working for a couple of days.”