Rethinking Recommendation Paradigms: From Pipelines to Agentic Recommender Systems

TL;DR

This paper introduces AgenticRS, a self-evolving recommender system framework that reorganizes traditional static pipelines into modular, evaluable agents capable of independent and collective evolution.

Contribution

It proposes a novel agent-based architecture for recommender systems, enabling self-evolution through reinforcement learning and large language model techniques.

Findings

AgenticRS modularizes recommender components as evaluable agents.

Self-evolution mechanisms include reinforcement learning and LLM-based architecture generation.

Layered reward design couples local agent optimization with global system objectives.

Abstract

Large-scale industrial recommenders typically use a fixed multi-stage pipeline (recall, ranking, re-ranking) and have progressed from collaborative filtering to deep and large pre-trained models. However, both multi-stage and so-called One Model designs remain essentially static: models are black boxes, and system improvement relies on manual hypotheses and engineering, which is hard to scale under heterogeneous data and multi-objective business constraints. We propose an Agentic Recommender System (AgenticRS) that reorganizes key modules as agents. Modules are promoted to agents only when they form a functionally closed loop, can be independently evaluated, and possess an evolvable decision space. For model agents, we outline two self-evolution mechanisms: reinforcement learning style optimization in well-defined action spaces, and large language model based generation and selection of new architectures and training schemes in open-ended design spaces. We further distinguish individual evolution of single agents from compositional evolution over how multiple agents are selected and connected, and use a layered inner and outer reward design to couple local optimization with global objectives. This provides a concise blueprint for turning static pipelines into self-evolving agentic recommender systems.