Explore Microsoft products and services and support for your home or business. Shop Microsoft 365, Copilot, Teams, Xbox, Windows, Azure, Surface and more. Backed by Menlo.
About the role
Learn how people work today: what they are trying to get done, what they have to keep track of, where they hesitate, and what they do when a tool does not fit. Then design better ways for them to work with agents and the AI capability around them. This includes permissions, consent, delegation, revocation, and clear limits on what an agent may do and on whose behalf. Design so that people can see when an agent is at work and decide for themselves how far to rely on it.
More about this role
Learn how people work today: what they are trying to get done, what they have to keep track of, where they hesitate, and what they do when a tool does not fit. Then design better ways for them to work with agents and the AI capability around them. This includes permissions, consent, delegation, revocation, and clear limits on what an agent may do and on whose behalf. Design so that people can see when an agent is at work and decide for themselves how far to rely on it. This covers how an agent announces itself and its work, how uncertainty is expressed, when escalation happens and when it does not, and what a review step must show for a person to be able to disagree with it. Design for supervision of the system while it is running. Someone overseeing several agents needs to work out which agent is doing what, why it did it, what it depends on, how far it has got, and when to step in. Design for the days it goes wrong. This covers override, rollback, correction, and how someone keeps working when a system is down or confidently mistaken. Set the quality bar for experiences and prove it with users. Included in this is task success, whether people could follow what happened, whether...
Browse similar: AI jobs · AI startup jobs · Startup jobs