nanoMuse: An Open-Source Personal Agent for Every Device You Own
THE QUICK READ
The report defines a personal agent through five questions and three horizons, then analyzes Meta's Muse using the company's public record and a copy of its production prompt. The authors present nanoMuse as an open-source GPL-3.0 counterpart: one agent across a person's devices, sharing a conversation over a self-hostable relay, routing actions through a Sentinel, keeping memory as readable files, and letting users choose the model.
KEY POINTS
- The authors frame the personal agent as software rather than a model, and describe Muse as closed and confined to one vendor's cloud in one country.
- nanoMuse is proposed with screen control on phone and computer, a shared conversation relay, Sentinel action review, file-based memory, and user-selected models.
- Size and cost are given only as estimates; provenance-aware memory, an evaluation suite for device actions, and an open model are listed as roadmap items.
READING LIMITS
Reading only the title and abstract leaves unclear how the authors reconstructed Muse from public records, how effective the Sentinel is in practice, and how the estimates and roadmap items were validated.
Source abstract [EN]
Assistants from 2011 answered and waited, and agents from 2023 did a task and stopped. In September 2026 Meta's Muse showed an agent for one person, with accounts, devices, memory and a conversation that lasts, closed, in a vendor's cloud, in one country. Such an agent is expected to act on a person's accounts and devices, remember them across weeks, speak first when it is worth it, and answer for what it did. It is a kind of software, not a model, and until now had no open counterpart. This report defines the personal agent in five questions and three horizons. It reads how Muse is built from Meta's public record and a copy of its production prompt, each statement marked by its source. It then presents nanoMuse, the open-source counterpart under the GPL-3.0, one agent on every device a person owns, with hands on the phone's screen and the computer's. They share one conversation over a relay anyone can run; every action goes through a Sentinel, memory is files the person can read, and the model is their choice. Its size and cost are given as estimates. What is open, memory with provenance, an evaluation suite for the hands and an open model for them, is set out as a roadmap.
AI summaries use public titles and abstracts only. Research claims belong to the authors; consult the original paper.