AI is beginning to do more than answer our questions. It can book the trip, send the message, move the money, write the code, negotiate with another system-and increasingly act in our name.
That changes the problem.
Once software can create real consequences on our behalf, it is no longer enough to ask whether an AI system is capable or safe. We also have to ask: Who gave it authority? What exactly was delegated? What happens when one agent hands work to another? And why should a bank, employer, platform, or other institution recognize that authority at all?
When Software Represents Us follows that problem from the individual agent outward. It begins with the shift from assistant to actor to representative, then examines the architecture that starts to matter once software acts for people and organizations: identity, permission, authority, delegation, system-level risk, institutional boundaries, and federation.
The book's central argument is simple but consequential: identity is not authority, permission is not a mandate, and individually safe agents can still create unsafe systems.
Drawing on real systems, emerging agent infrastructure, identity architecture, finance, and institutional design, Johnny Kao shows why many problems now treated separately are beginning to converge as general-purpose agents become capable of representing us.
This is not a coding manual, a prediction of artificial superintelligence, or a proposal for one universal governance protocol. It is a practical framework for understanding the institutional architecture that will be required when software stops merely assisting us and begins acting on our behalf.
