Saved in:
Bibliographic Details
Main Authors: She, Yifei, Zhang, Ping, Liu, He, Jia, Yanmin, Jing, Yang, Liu, Zijun, Sun, Peng, Li, Xiangbin, Hu, Xiaohe
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.16262
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912773218238464
author She, Yifei
Zhang, Ping
Liu, He
Jia, Yanmin
Jing, Yang
Liu, Zijun
Sun, Peng
Li, Xiangbin
Hu, Xiaohe
author_facet She, Yifei
Zhang, Ping
Liu, He
Jia, Yanmin
Jing, Yang
Liu, Zijun
Sun, Peng
Li, Xiangbin
Hu, Xiaohe
contents Real-world agentic tasks, unlike synchronous Markov Decision Processes (MDPs), often involve non-blocking actions with variable latencies, creating a fundamental \textit{Temporal Gap} between action initiation and completion. Existing environment-side solutions, such as blocking wrappers or frequent polling, either limit scalability or dilute the agent's context window with redundant observations. In this work, we propose an \textbf{Agent-side Approach} that empowers Large Language Models (LLMs) to actively align their \textit{Cognitive Timeline} with the physical world. By extending the Code-as-Action paradigm to the temporal domain, agents utilize semantic priors and In-Context Learning (ICL) to predict precise waiting durations (\texttt{time.sleep(t)}), effectively synchronizing with asynchronous environment without exhaustive checking. Experiments in a simulated Kubernetes cluster demonstrate that agents can precisely calibrate their internal clocks to minimize both query overhead and execution latency, validating that temporal awareness is a learnable capability essential for autonomous evolution in open-ended environments.
format Preprint
id arxiv_https___arxiv_org_abs_2512_16262
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning to Wait: Synchronizing Agents with the Physical World
She, Yifei
Zhang, Ping
Liu, He
Jia, Yanmin
Jing, Yang
Liu, Zijun
Sun, Peng
Li, Xiangbin
Hu, Xiaohe
Artificial Intelligence
Real-world agentic tasks, unlike synchronous Markov Decision Processes (MDPs), often involve non-blocking actions with variable latencies, creating a fundamental \textit{Temporal Gap} between action initiation and completion. Existing environment-side solutions, such as blocking wrappers or frequent polling, either limit scalability or dilute the agent's context window with redundant observations. In this work, we propose an \textbf{Agent-side Approach} that empowers Large Language Models (LLMs) to actively align their \textit{Cognitive Timeline} with the physical world. By extending the Code-as-Action paradigm to the temporal domain, agents utilize semantic priors and In-Context Learning (ICL) to predict precise waiting durations (\texttt{time.sleep(t)}), effectively synchronizing with asynchronous environment without exhaustive checking. Experiments in a simulated Kubernetes cluster demonstrate that agents can precisely calibrate their internal clocks to minimize both query overhead and execution latency, validating that temporal awareness is a learnable capability essential for autonomous evolution in open-ended environments.
title Learning to Wait: Synchronizing Agents with the Physical World
topic Artificial Intelligence
url https://arxiv.org/abs/2512.16262