CRMWeaver: Building Powerful Business Agent via Agentic RL and Shared Memories

2025-10-29

AI Summary

Commercial agents face challenges in handling heterogeneous tasks (e.g., statistical queries, knowledge-based QA) and modeling relationships across multi-source data in complex business environments. Method: This paper proposes an intelligent agent framework integrating reinforcement learning (RL) with a shared memory mechanism. We introduce an agentic RL training paradigm driven by synthetic data, combining an LLM-based tool-calling architecture with a reusable shared memory module to enable cross-task experience transfer and improve reasoning consistency. Contribution/Results: Our key innovation lies in embedding the memory mechanism directly into the RL feedback loop, substantially enhancing generalization to unseen tasks. Evaluated on the CRMArena-Pro benchmark, our lightweight model achieves state-of-the-art performance in both B2B and B2C scenarios, demonstrating effectiveness, robustness, and industrial scalability.

Technology Category

• Multiagent Systems: Multiagent Learning
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Machine Learning: Reinforcement Learning

Application Category

• 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
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions

Abstract

Recent years have witnessed the rapid development of LLM-based agents, which shed light on using language agents to solve complex real-world problems. A prominent application lies in business agents, which interact with databases and internal knowledge bases via tool calls to fulfill diverse user requirements. However, this domain is characterized by intricate data relationships and a wide range of heterogeneous tasks, from statistical data queries to knowledge-based question-answering. To address these challenges, we propose CRMWeaver, a novel approach that enhances business agents in such complex settings. To acclimate the agentic model to intricate business environments, we employ a synthesis data generation and RL-based paradigm during training, which significantly improves the model's ability to handle complex data and varied tasks. During inference, a shared memories mechanism is introduced, prompting the agent to learn from task guidelines in similar problems, thereby further boosting its effectiveness and generalization, especially in unseen scenarios. We validate the efficacy of our approach on the CRMArena-Pro dataset, where our lightweight model achieves competitive results in both B2B and B2C business scenarios, underscoring its practical value for real-world applications.

Problem

• Enhancing business agents for complex data relationships
• Improving agent performance in heterogeneous business tasks
• Boosting generalization in unseen business scenarios

Innovation

• Agentic RL training for complex business environments
• Shared memories mechanism for task generalization
• Synthesis data generation to handle varied tasks