arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.00495cs.AI
收藏

AI 运行时基础设施

AI Runtime Infrastructure

Christopher Cruz

AI总结:

本文提出AI运行时基础设施,通过在模型与应用之间引入执行层,主动监控和干预智能体行为,优化任务成功率、延迟、令牌效率、可靠性和安全性。

摘要:

我们介绍了AI运行时基础设施,一种在模型和应用之间运作的独立执行层,能够主动观察、推理并干预智能体行为,以优化任务成功、延迟、令牌效率、可靠性和安全性。与模型级优化或被动日志系统不同,运行时基础设施将执行本身视为优化表面,使在长时间跨度的智能体工作流中实现自适应内存管理、故障检测、恢复和策略执行。

英文摘要:

We introduce AI Runtime Infrastructure, a distinct execution-time layer that operates above the model and below the application, actively observing, reasoning over, and intervening in agent behavior to optimize task success, latency, token efficiency, reliability, and safety while the agent is running. Unlike model-level optimizations or passive logging systems, runtime infrastructure treats execution itself as an optimization surface, enabling adaptive memory management, failure detection, recovery, and policy enforcement over long-horizon agent workflows.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑