MACS: A Hybrid Multi-Agent Framework for Reliable Conversational E-Commerce Recommendation

2026-08-14

AI Summary

This study addresses the challenges of reliability and hard constraint adherence in fixed-catalog conversational recommendation by proposing a hybrid multi-agent framework. The approach innovatively integrates the semantic understanding of large language models with deterministic agents, employing a session preference persistence layer and a progressive relaxation mechanism to effectively prevent constraint drift and ensure cross-turn consistency. Experimental results demonstrate that the system achieves an 87.1% single-turn pass rate with perfect brand compliance, alongside a 72% multi-turn pass@5 score. Notably, it attains 100% accuracy in constraint inversion and accumulation scenarios, significantly enhancing both regulatory compliance and stability in e-commerce recommendation systems.

Abstract

Conversational recommendation for e-commerce is increasingly mediated by large language models (LLMs), yet many real-world deployments operate under a stricter requirement: recommendations must be drawn only from a merchant's fixed catalog, without web search or unsupported product claims. In this setting, the main challenge is reliability under hard constraints: the system must satisfy user requirements, remain grounded in available inventory, and preserve preferences across multiple conversational turns. We present MACS (Multi-Agent Commerce System), a hybrid multi-agent framework for reliable conversational recommendation in fixed-catalog settings. MACS uses LLMs for language-facing tasks such as interpreting user requests, eliciting preferences, and generating responses, while correctness-critical operations, including product retrieval, hard-constraint filtering, brand exclusion, and progressive relaxation, are executed deterministically by the merchant agent. A session-persistent preference layer tracks constraints across turns, enabling consistent handling of budget overwrites and exclusion reversals. On a 140-query single-turn benchmark, MACS achieves the highest pass rate (87.1%) and perfect brand compliance (1.000). On a 10-scenario multi-turn benchmark, MACS achieves the strongest macro Pass@5 (72% vs. 56% GPT+Catalog / 52% Gemini+Catalog) with zero constraint drift. The advantage is sharpest on exclusion reversal (100% vs. 20% / 0%) and constraint accumulation (100% vs. 60% / 40%). Mean judged response quality is similar across systems (0.751 vs. 0.736). These results suggest that hybrid architectures combining deterministic constraint enforcement with session-persistent preference tracking provide stronger reliability-oriented performance than catalog-bound prompt-only baselines in the fixed-catalog merchant setting.

Problem

• Conversational Recommendation
• Fixed-Catalog
• Reliability
• Hard Constraints
• Multi-turn Dialogue

Innovation

• Hybrid Multi-Agent Framework
• Deterministic Constraint Enforcement
• Session-Persistent Preference Layer
• Fixed-Catalog Recommendation
• Progressive Relaxation