Agentic Edge AI

Di Wu; Jiawei Luo; Yizhou Han; Weisong Shi; Blesson Varghese

Agentic Edge AI

Di Wu
, Jiawei Luo
, Yizhou Han
, Weisong Shi
, Blesson Varghese

School of Computer Science

Research output: Working paper › Preprint

Abstract

This article introduces Agentic Edge AI, a new research direction for the Edge AI community. Rather than optimizing predefined edge tasks, it enables goal-oriented agents to synthesize, execute, and refine edge workflows from high-level user goals. We discuss the opportunities enabled by this vision, illustrate it through a motivating example of a smart factory, and identify the key design considerations and research challenges for building Agentic Edge AI systems.

Original language	English
Publication status	Submitted - 5 Aug 2026

Keywords

Agentic Edge AI
Edge Computing
Generative AI Agents

Handle.net

Persistent link to portal record

Cite this

@techreport{6d09a941e2ed45d8803642c9a8cf7371,

title = "Agentic Edge AI",

abstract = "This article introduces Agentic Edge AI, a new research direction for the Edge AI community. Rather than optimizing predefined edge tasks, it enables goal-oriented agents to synthesize, execute, and refine edge workflows from high-level user goals. We discuss the opportunities enabled by this vision, illustrate it through a motivating example of a smart factory, and identify the key design considerations and research challenges for building Agentic Edge AI systems.",

keywords = "Agentic Edge AI, Edge Computing, Generative AI Agents",

author = "Di Wu and Jiawei Luo and Yizhou Han and Weisong Shi and Blesson Varghese",

year = "2026",

month = aug,

day = "5",

language = "English",

type = "WorkingPaper",

}

TY - UNPB

T1 - Agentic Edge AI

AU - Wu, Di

AU - Luo, Jiawei

AU - Han, Yizhou

AU - Shi, Weisong

AU - Varghese, Blesson

PY - 2026/8/5

Y1 - 2026/8/5

N2 - This article introduces Agentic Edge AI, a new research direction for the Edge AI community. Rather than optimizing predefined edge tasks, it enables goal-oriented agents to synthesize, execute, and refine edge workflows from high-level user goals. We discuss the opportunities enabled by this vision, illustrate it through a motivating example of a smart factory, and identify the key design considerations and research challenges for building Agentic Edge AI systems.

AB - This article introduces Agentic Edge AI, a new research direction for the Edge AI community. Rather than optimizing predefined edge tasks, it enables goal-oriented agents to synthesize, execute, and refine edge workflows from high-level user goals. We discuss the opportunities enabled by this vision, illustrate it through a motivating example of a smart factory, and identify the key design considerations and research challenges for building Agentic Edge AI systems.

KW - Agentic Edge AI

KW - Edge Computing

KW - Generative AI Agents

UR - https://www.researchgate.net/publication/411632031_Agentic_Edge_AI

M3 - Preprint

BT - Agentic Edge AI

ER -

Agentic Edge AI

Abstract

Keywords

Handle.net

Other files and links

Fingerprint

Cite this