Asynchronous Remote Interaction and Conversational Interrupt Handling for Long-Running Agentic Tasks
Abstract
Tasks are managed through a decoupled interaction and execution framework. An execution environment on a primary device performs a task and detects an interrupt state when a decision node is reached that lacks parameters or confirmation. A natural language query object is serialized and routed via a cloud messaging service or chat to an authenticated secondary device. Remote input is captured at the secondary device and transmitted back to the primary device. The input is parsed and injected into the execution runtime, allowing the task to resume from the point of interruption. This process enables the extension of decision-making logic to remote clients, including mobile devices, speakers, and wearable electronics, without requiring full remote desktop access or manual navigation of the primary host interface.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Liu, Yijo; Shin, Yeun Su; Chan, Clifford Tse Yan; Walsh, Shadia; Duraiswami, Lily; Borg, Carl Magnus; Morris, Tiffany Deonne; and Krishna, Golden Gopal, "Asynchronous Remote Interaction and Conversational Interrupt Handling for Long-Running Agentic Tasks", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10064