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

Researchers developed ProductResearch, a multi-agent AI framework that creates synthetic training data to improve e-commerce shopping agents. The system uses multiple AI agents to generate comprehensive product research trajectories, with experiments showing a compact model fine-tuned on this synthetic data significantly outperforming base models in shopping assistance tasks.

• →ProductResearch framework generates synthetic training trajectories using User, Supervisor, and Research agents working collaboratively.
• →The system addresses limitations in existing LLM-based e-commerce agents that lack interaction depth and contextual understanding.
• →Synthetic trajectory distillation converts multi-agent interactions into single-role training examples for effective model fine-tuning.
• →Experiments demonstrate substantial improvements in response comprehensiveness and research depth for shopping queries.
• →The approach provides a scalable method for training LLM-based shopping assistants without requiring extensive real-world data.