Meta launches Muse personal AI agent at ai.meta.com/muse

Meta released Muse as an interface-layer agent that accepts natural language commands and executes tasks across Meta products. The deployment follows Meta's prior Llama model releases and builds on internal agent frameworks tested in 2024. Primary documentation at ai.meta.com/muse lists supported actions including content generation, scheduling, and profile management without external API calls.

No public benchmark scores or latency figures appear in the release. Hacker News discussion reached 516 points with 554 comments focused on integration depth versus OpenAI and Anthropic agent offerings. The absence of reported accuracy metrics or user retention data leaves operational claims unverified against external datasets.

Muse extends Meta's pattern of embedding Llama-derived models into consumer surfaces after the 2023-2024 Llama 3 and Llama 3.1 rollouts. This reduces friction for daily AI use but inherits Llama's documented limitations in long-context reasoning and tool-use consistency. Operational impact depends on whether Meta routes agent traffic through existing Instagram and WhatsApp sessions at scale.

Next steps include potential expansion of Muse actions to third-party services and tighter coupling with Ray-Ban Meta glasses hardware announced in prior quarters.