AgenticRS-Architecture: System Design for Agentic Recommender Systems

2026-03-27

AI Summary

This work proposes AutoModel, a novel architecture that reimagines recommender systems as a multi-agent system endowed with long-term memory and self-optimization capabilities, addressing the limitations of conventional fixed recall-and-ranking pipelines that hinder cross-module collaboration and autonomous evolution. AutoModel integrates three core dimensions—models, features, and resources—and achieves synergistic co-evolution through a coordination layer that aligns local automation with global objectives via knowledge sharing and code generation mechanisms. A key component, AutoTrain, enables automatic model reproduction from academic papers and supports large-scale training. Experimental results demonstrate that the proposed approach substantially reduces manual migration costs and exhibits broad applicability across AI-driven systems such as search and online advertising.

Abstract

AutoModel is an agent based architecture for the full lifecycle of industrial recommender systems. Instead of a fixed recall and ranking pipeline, AutoModel organizes recommendation as a set of interacting evolution agents with long term memory and self improvement capability. We instantiate three core agents along the axes of models, features, and resources: AutoTrain for model design and training, AutoFeature for data analysis and feature evolution, and AutoPerf for performance, deployment, and online experimentation. A shared coordination and knowledge layer connects these agents and records decisions, configurations, and outcomes. Through a case study of a module called paper autotrain, we show how AutoTrain automates paper driven model reproduction by closing the loop from method parsing to code generation, large scale training, and offline comparison, reducing manual effort for method transfer. AutoModel enables locally automated yet globally aligned evolution of large scale recommender systems and can be generalized to other AI systems such as search and advertising.

Problem

• recommender systems
• automation
• agent-based architecture
• model evolution
• feature engineering

Innovation

• Agentic Recommender Systems
• AutoModel
• Evolution Agents
• Self-improving AI
• Automated Model Reproduction