The line that mattered on the call is the conversational-assistant story — and Microsoft's grant US11947902B1 (“Efficient multi-turn generative AI model suggested message generation,” issued 2024-04-02) is the IP under it. Assigned to Microsoft Technology Licensing, LLC and classified CPC G06F 40/40 and G06N 3/0475, it covers generating suggested messages efficiently across multiple conversational turns.
Management sold assistants that carry a conversation — and the word “efficient” in the title is the financial tell. Multi-turn generation is expensive because context grows with every turn; doing it efficiently is the difference between an assistant that's cheap to run and one that quietly burns margin per conversation.
“Systems and methods for using a generative artificial intelligence (AI) model using a multi-turn process to generate a suggested draft reply to a selected message.”— U.S. Patent No. 11,947,902 source
Microsoft discloses AI revenue through cloud and productivity segments and references assistant products in commentary without isolating per-feature economics. The grant is the granular evidence under that story: dated 2024, owned, and pointed at the cost and capability of the multi-turn experience that the revenue pitch depends on.
Read the claim, not the marketing: a grant is invention and ownership, not a revenue figure, and we attribute none. It also doesn't prove a named product ships this exact method. What it documents is dated IP behind a monetized, conversational capability — with efficiency, not just capability, claimed.
For the calls desk, the discipline is to notice when a patent title quietly answers the cost question management dodged. “Efficient multi-turn” generation is a primary document suggesting Microsoft was engineering the unit economics of conversation, not just the experience.