Abstract

Myanmar's emerging secondhand and local fashion merchants operate predominantly through Facebook and Viber, lacking automated tools for order management, AI-assisted discovery, or consumer trust verification. This paper presents a survey-based requirements analysis and a corresponding Model Context Protocol (MCP)-driven multi-agent enterprise architecture for a dedicated marketplace platform targeting this segment. A structured needs-assessment survey of 15 merchants identified operational pain points and ranked 16 platform capabilities on a five-point Likert scale. Features scoring 3.85 or above on the survey were adopted as active requirements, producing eleven survey-qualified capabilities. A twelfth — AI condition grading — was added independent of its survey score, since it serves as a trust signal rather than a popularity measure. Together, these twelve capabilities map across five MCP tool servers: analytics, order-manager, recommendation, chat-assistant, and product-listing. The proposed five-layer architecture spans a React Native/Next.js client layer, a LangGraph multi-agent orchestration layer coordinated via the Agent-to-Agent (A2A) protocol, the MCP tool layer, a MongoDB/Redis data infrastructure layer, and an observability and monitoring layer. Trust is the biggest factor driving purchases in Myanmar social commerce. To address this, the architecture builds a trust layer — based on signaling theory — into its orchestration and MCP tool layers, using AI condition grading, seller reputation scores, and verified-seller badges. This approach offers a repeatable model for building AI-first marketplaces in similar emerging markets.

References

• DataReportal. 2024. Digital 2024: Myanmar — Global Digital Insights. Retrieved from https://datareportal.com/reports/digital-2024-myanmar
• Aung, T., Liana, S. R., Htet, A., and Bhaumik, A. 2024. Analyzing the determinants of consumer buying behavior on Facebook Marketplace in Myanmar: a quantitative approach. J. Electr. Syst. 20, 9s (2024), 1142–1152. Retrieved from https://journal.esrgroups.org/jes/article/view/4473
• Anthropic. 2024. Model Context Protocol Specification. Retrieved from https://modelcontextprotocol.io
• Negash, Y. T. and Akhbar, T. 2024. Building consumer trust in secondhand fashion: a signaling theory perspective on how consumer orientation and environmental awareness shape engagement. Clean. Responsible Consum. 14 (2024), 100211. DOI:https://doi.org/10.1016/j.clrc.2024.100211
• Ricci, F., Rokach, L., and Shapira, B. (Eds.) 2015. Recommender Systems Handbook. 2nd ed. Springer, New York.
• Bae, Y., Choi, J., Gantumur, M., and Kim, N. 2022. Technology-based strategies for online secondhand platforms promoting sustainable retailing. Sustainability 14, 6 (2022), 3259. DOI:https://doi.org/10.3390/su14063259
• Economic Research Institute for ASEAN and East Asia (ERIA). 2021. Digital Economy of Myanmar. In Digital Economy Integration in ASEAN, chap. 6. ERIA, Jakarta, Indonesia. Retrieved from https://www.eria.org/uploads/6_ch_6-Myanmar.pdf
• Kyi, N. P. W. 2024. An empirical study of information search and alternative evaluation, purchase and post purchase affecting customer satisfaction and e-commerce business development in Myanmar. M.S. thesis, Bangkok Univ., Bangkok, Thailand.
• Lita, R. P., Meuthia, M., Rahmi, D. Y., and Syafrida, M. F. 2024. Does social presence on social commerce platform attract buying intention of Indonesian local food? J. Appl. Eng. Technol. Sci. 6, 1 (2024), 174–191. DOI:https://doi.org/10.37385/jaets.v6i1.5852
• Krishnan, N. K. 2025. Advancing multi-agent systems through Model Context Protocol: architecture, implementation, and applications. Preprint arXiv:2504.21030. Retrieved from https://arxiv.org/abs/2504.21030
• Chen, K., Sun, Y., Keung, J., Mao, Z., and Ma, X. 2026. Understanding how enterprises adopt the Model Context Protocol for LLM-driven software engineering. Preprint arXiv:2606.09182. Retrieved from https://arxiv.org/abs/2606.09182
• Wooldridge, M. 2009. An Introduction to MultiAgent Systems. 2nd ed. Wiley, Chichester, UK.
• Google. 2024. Agent-to-Agent (A2A) Protocol Documentation. Retrieved from https://google.github.io/A2A/
• Witkin, B. R. and Altschuld, J. W. 1995. Planning and Conducting Needs Assessments: A Practical Guide. SAGE Publications, Thousand Oaks, CA.
• LangChain Inc. 2024. LangGraph Documentation. Retrieved from https://langchain-ai.github.io/langgraph/
• LangChain Inc. 2024. LangChain Documentation. Retrieved from https://python.langchain.com/
• Qdrant Solutions GmbH. 2024. Qdrant Documentation. Retrieved from https://qdrant.tech/documentation/
• Langfuse GmbH. 2024. LangFuse Documentation. Retrieved from https://langfuse.com/docs
• Databricks Inc. 2024. MLflow Documentation. Retrieved from https://mlflow.org/docs/
• He, K., Zhang, X., Ren, S., and Sun, J. 2016. Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 770–778.
• Neo4j Inc. 2024. Neo4j Documentation. Retrieved from https://neo4j.com/docs/