[論文レビュー] The Next Paradigm Is User-Centric Agent, Not Platform-Centric Service

Luankang Zhang, Hang Lv+9 more|arXiv (Cornell University)|Feb 17, 2026

AI in Service Interactions被引用数 0

ひとこと要約

Argues for shifting from platform-centric services to user-centric agents with a device–cloud pipeline and user governance to prioritize user utility and privacy.

ABSTRACT

Modern digital services have evolved into indispensable tools, driving the present large-scale information systems. Yet, the prevailing platform-centric model, where services are optimized for platform-driven metrics such as engagement and conversion, often fails to align with users' true needs. While platform technologies have advanced significantly-especially with the integration of large language models (LLMs)-we argue that improvements in platform service quality do not necessarily translate to genuine user benefit. Instead, platform-centric services prioritize provider objectives over user welfare, resulting in conflicts against user interests. This paper argues that the future of digital services should shift from a platform-centric to a user-centric agent. These user-centric agents prioritize privacy, align with user-defined goals, and grant users control over their preferences and actions. With advancements in LLMs and on-device intelligence, the realization of this vision is now feasible. This paper explores the opportunities and challenges in transitioning to user-centric intelligence, presents a practical device-cloud pipeline for its implementation, and discusses the necessary governance and ecosystem structures for its adoption.

研究の動機と目的

• Motivate the shift from platform-centric to user-centric intelligence in digital services.
• Identify structural bottlenecks in platform-centric models (fragmented context, limited execution boundaries, misaligned incentives).
• Propose a concrete device–cloud pipeline to implement user-centric agents.
• Discuss governance, ecosystem contracts, and economic implications for adoption.

提案手法

• Analyze limitations of platform-centric services and argue for user-centric agents driven by user-defined goals and constraints.
• Propose an edge–cloud collaborative pipeline with an on-device personal agent that reasons locally and queries external inventories via cloud services.
• Describe a three-stage on-device lifecycle: Perception and Memory, Intent Decomposition and Planning, Execution and Evolution.
• Outline cloud role as an agentic service layer providing context-aware reasoning, adaptive strategy modulation, and explainable generation.
• Address governance: ownership, portability of memory, and contracts that balance user sovereignty with platform cooperation.

実験結果

リサーチクエスチョン

• RQ1 What structural bottlenecks in current platform-centric services prevent true user utility?
• RQ2 How can a device–cloud pipeline realize a practical user-centric agent while maintaining privacy and control?
• RQ3 What governance and ecosystem contracts are needed to enable user-governed agents without reinforcing platform-centric capture?
• RQ4 How can on-device perception, planning, and execution be integrated with cloud-powered external inventories?
• RQ5 What changes are required in data, model, and inference strategies to support user-centric intelligence?

主な発見

• A holistic, user-centric agent can address fragmented context by keeping signals on-device and aggregating cross-app data locally.
• Intent fulfillment across services requires end-to-end planning beyond single-platform boundaries, aligned to user constraints rather than platform metrics.
• An edge–cloud pipeline enables on-device sovereignty while leveraging cloud access to external inventories and tools.
• Governance mechanisms (portability of memory, standardized schemas, auditability, revenue-sharing contracts) are essential to prevent platform-centric capture.
• On-device memory, structured planning, and verifiable execution create a self-improving loop that strengthens personalization without exporting raw private data.
• The paradigm shift is feasible due to advances in LLMs, on-device intelligence, and architecture that combines device control with cloud-enabled execution.