Meta Muse and the Next Frontier of Agentic AI: Aligning Autonomy with Intent

This article examines Meta’s agentic AI strategy through Muse, its personal agent designed to draw on past interactions, connected applications, and, eventually, the user’s physical environment. As Meta extends Muse to businesses and developers through the Meta Enterprise Platform, the paper considers a central enterprise challenge: intent drift, in which an agent operates within its permissions but diverges from the user’s intended objective. It traces how this risk emerges across an agent’s context, memory, and actions, and outlines five trust boundaries that enterprises can use to assess an agent’s governance from an intent alignment perspective before expanding autonomy.