Search

Cognitive Dialogue: Modular LLM Agents for Context-Aware and Multimodal Customer Service

Author(s)

Shaojue Yan*

Affiliation(s)

Xi’an Jiaotong-Liverpool University, Suzhou, China *Corresponding Author

Abstract

As companies expand, the need for effective customer service has grown, putting pressure on traditional methods due to rising costs and resource constraints. Shen (2025) explained that recent developments in Artificial Intelligence (AI), particularly with Large Language Models (LLMs), provide promising solutions by offering automated, scalable, and cost-efficient customer support. This paper examines the role of LLMs in intelligent customer service systems, highlighting their ability to hold dynamic conversations, provide multi-language support, and deliver personalized service. By incorporating advanced techniques such as Prompt Engineering and Retrieval-Augmented Generation (RAG), we present a model designed to improve LLM performance, enhancing efficiency, accuracy, and customer satisfaction. Furthermore, we compare the evolution of intelligent customer service systems, contrasting rule-based and deep learning-based models. Our findings suggest that LLM-driven systems can significantly boost service efficiency, lower operational costs, and improve user experiences.

Keywords

LLMs; RAG; Multimodal Interfaces; Human-Computer Collaboration

References

[1] Shen, L., Yang, Q., Zheng, Y., & Li, M. (2025). AutoIOT: LLM-Driven Automated Natural Language Programming for AIoT Applications. arXiv preprint arXiv:2503.05346. [2] Chkirbene, Z., Hamila, R., Gouissem, A., & Devrim, U. (2024, December). Large Language Models (LLM) in Industry: A Survey of Applications, Challenges, and Trends. In 2024 IEEE 21st International Conference on Smart Communities: Improving Quality of Life using AI, Robotics and IoT (HONET) (pp. 229-234). IEEE [3] Sowmiya, R., Revathi, P., Ragunath, D., Gokila, P., & Kalaivani, T. (2024, October). Multi-Modal LLM Driven Computer Interface. In 2024 8th International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC) (pp. 484-489). IEEE. [4] Shah, C., White, R. W., Andersen, R., Buscher, G., Counts, S., Das, S. S. S., ... & Yang, L. (2023). Using large language models to generate, validate, and apply user intent taxonomies. arXiv preprint arXiv:2309.13063. [5] Friha, O., Ferrag, M. A., Kantarci, B., Cakmak, B., Ozgun, A., & Ghoualmi-Zine, N. (2024). llm-based edge intelligence: A comprehensive survey on architectures, applications, security and trustworthiness. IEEE Open Journal of the Communications Society. [6] Qin, J., Wu, J., Chen, W., Ren, Y., Li, H., Wu, H., ... & Wen, S. (2024). DiffusionGPT: LLM-driven text-to-image generation system. arXiv preprint arXiv:2401.10061. [7] Yao, Y., Duan, J., Xu, K., Cai, Y., Sun, Z., & Zhang, Y. (2024). A survey on large language model (llm) security and privacy: The good, the bad, and the ugly. High-Confidence Computing, 100211.