LLMs Working in Harmony: A Survey on the Technological Aspects of Building Effective LLM-Based Multi Agent Systems

TL;DR

This survey tackles the problem of building effective LLM-based multi-agent systems by examining four core technological pillars: Architecture, Memory, Planning, and Technologies/Frameworks. It surveys representative approaches such as Mixture-of-Agents for collaborative architecture and ReAct for integrated reasoning and action, highlighting their strengths and limitations. The findings offer a practical roadmap: MoA and ReAct stand out for coordinating agents and planning under dynamic conditions, while memory strategies (e.g., RAG, VecDB, symbolic memory) enable persistent, knowledge-rich interactions; the choice of framework is task-dependent, underscoring the need for adaptable, scalable tooling. The practical impact lies in guiding researchers and engineers toward scalable, robust MAS designs that balance performance, cost, and reliability in real-world settings.

Abstract

This survey investigates foundational technologies essential for developing effective Large Language Model (LLM)-based multi-agent systems. Aiming to answer how best to optimize these systems for collaborative, dynamic environments, we focus on four critical areas: Architecture, Memory, Planning, and Technologies/Frameworks. By analyzing recent advancements and their limitations - such as scalability, real-time response challenges, and agent coordination constraints, we provide a detailed view of the technological landscape. Frameworks like the Mixture of Agents architecture and the ReAct planning model exemplify current innovations, showcasing improvements in role assignment and decision-making. This review synthesizes key strengths and persistent challenges, offering practical recommendations to enhance system scalability, agent collaboration, and adaptability. Our findings provide a roadmap for future research, supporting the creation of robust, efficient multi-agent systems that advance both individual agent performance and collective system resilience.

Figures (1)

• Figure 1: Overview of Survey Methodology