ProductResearch: Training E-Commerce Deep Research Agents via Multi-Agent Synthetic Trajectory Distillation

2026-02-27

AI Summary

This work addresses the limitations of existing large language model agents in supporting complex product research tasks within e-commerce settings, where deep interaction and broad contextual understanding are often lacking. To bridge this gap, the authors propose a multi-agent collaborative framework: a user agent infers shopping intent, while a supervisor agent orchestrates research agents to generate high-fidelity, long-horizon tool-use trajectories. These collaborative trajectories are then distilled—through reflection and internalization—into single-role training samples. This approach introduces a novel multi-agent synthetic trajectory distillation mechanism that transforms intricate collaborative processes into scalable training data for a single model. A fine-tuned, compact mixture-of-experts (MoE) model trained on this distilled data significantly outperforms baseline systems in answer comprehensiveness, research depth, and perceived user utility, achieving performance comparable to state-of-the-art closed-source systems.

Technology Category

• Multiagent Systems: Teamwork
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Agentic search

Abstract

Large Language Model (LLM)-based agents show promise for e-commerce conversational shopping, yet existing implementations lack the interaction depth and contextual breadth required for complex product research. Meanwhile, the Deep Research paradigm, despite advancing information synthesis in web search, suffers from domain gaps when transferred to e-commerce. We propose ProductResearch, a multi-agent framework that synthesizes high-fidelity, long-horizon tool-use trajectories for training robust e-commerce shopping agents. The framework employs a User Agent to infer nuanced shopping intents from behavioral histories, and a Supervisor Agent that orchestrates iterative collaboration with a Research Agent to generate synthetic trajectories culminating in comprehensive, insightful product research reports. These trajectories are rigorously filtered and distilled through a reflective internalization process that consolidates multi-agent supervisory interactions into coherent single-role training examples, enabling effective fine-tuning of LLM agents for complex shopping inquiries. Extensive experiments show that a compact MoE model fine-tuned on our synthetic data achieves substantial improvements over its base model in response comprehensiveness, research depth, and user-perceived utility, approaching the performance of frontier proprietary deep research systems and establishing multi-agent synthetic trajectory training as an effective and scalable paradigm for enhancing LLM-based shopping assistance.

Problem

• e-commerce
• deep research
• large language models
• product research
• conversational shopping

Innovation

• multi-agent framework
• synthetic trajectory distillation
• e-commerce deep research
• reflective internalization
• MoE fine-tuning