A new preprint on arXiv proposes PEAM, an on-device communication agent built around compact large language models. The system is designed to make autonomous decisions about mobile network configuration, a task that typically requires user intervention or remote processing. By keeping the decision-making on the smartphone, PEAM aims to offer a more personalized and efficient alternative to existing approaches.

The abstract notes that compact LLMs are already enabling autonomous on-device actions, such as mobile GUI agents. PEAM applies this capability to network settings, addressing scenarios where users need adaptive configuration. The paper positions the work as an extension of those emerging agent frameworks, though the full details of the implementation and evaluation are not available in the abstract alone.

As a single-source item, this summary reflects only the claims in the arXiv abstract. No independent validation or comparison with other methods is provided in the source.