"The answer is both — we've barely scratched the surface of all three monetization models"
Evidence from the Archive
Microsoft
Cursor hitting $300M ARR in 2 years -- a competitive challenge to Microsoft's Copilot
Microsoft Copilot across Office products (seat-based augmentation) alongside agent research projects
Oversees AI product strategy at Microsoft including Copilot and agent initiatives, previously CPO at Robinhood and VP at Google where she led Google Lens, Search, and AI Assistant -- giving her experience across consumer and enterprise AI at massive scale. Their core argument: Both models will coexist -- agents represent an evolution along a spectrum of autonomy, complexity, and asynchrony, not a binary replacement.
The evidence is specific: Microsoft Copilot across Office products (seat-based augmentation) alongside agent research projects. Furthermore, frontier Program: internal fake company where teams use cutting-edge agent tools to discover new work patterns. Deep research agent for work: asked it to analyze meeting attendees' views and suggest persuasion strategies.
In Aparna Chennapragada's own words: "We've just barely scratched the surface of whether you do seat monetization, usage like on tap, and then of course outcome-based stuff, outcome-based monetization. Hey, have you solved the problem for me and then I will pay you some fees. So all three to me are kind of like, great, but at least two out of three for a good product." (On the three coexisting monetization models for AI products.)