arXiv:2606.23049v2 Announce Type: replace Abstract: Phones are becoming an important execution surface for general-purpose agents, but training open models for reliable phone use remains difficult because the environment that matters at deployment, real devices running real apps, is slow, stateful, side-effectful, and hard to reset or verify, while scalable mock environments only approximate real behavior. We present PhoneBuddy, a training recipe and open-model line for agentic phone use that combines a real-app environment with a mock-app environment, PhoneWorld, which reconstructs runnable m
The increasing sophistication of large language models and the push for more autonomous AI applications are driving efforts to enable agents to interact with complex real-world environments like smartphones.
This development is crucial for expanding the capabilities of AI agents beyond simulated environments, allowing them to perform valuable actions on ubiquitous personal devices.
The ability to train open models for reliable agentic phone use will enable more practical and widespread deployment of AI agents in everyday tasks, blurring the lines between user and autonomous system.
- · AI agent developers
- · Smartphone manufacturers
- · Software developers (app automation)
- · Consumers (via advanced phone features)
- · Manual mobile task workers
- · Companies relying on repetitive digital human labor
AI agents gain the foundational ability to directly interface with and control smartphone applications.
This capability leads to a rapid proliferation of highly personalized and automated mobile AI assistants for various tasks.
The definition of phone 'use' shifts profoundly as a significant portion of interactions become agent-mediated, raising questions about data privacy and digital autonomy.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL