---
title: LLM-Based Multi-Agent Systems
url: https://www.emergentmind.com/topics/llm-based-multi-agent-systems-c730e815-6eb0-4219-a652-47653955a70d
type: topic
---

# LLM-Based Multi-Agent Systems

Large Language Model (LLM)-Based Multi-Agent Systems (MAS) are computational frameworks in which multiple autonomous agents, each powered by a large language model, collaborate to solve complex tasks, coordinate actions, or simulate sophisticated environments. These systems are characterized by dynamic inter-agent communication, emergent division of labor, and collective problem solving, and are deployed across domains ranging from creative generation and scientific research to cybersecurity, control engineering, and economic simulation. Recent research establishes taxonomies, formal models, and practical protocols that guide the architecture, coordination, evaluation, and safety of LLM-based MAS [2505.21116][2402.01680][2411.14033].

## 1. Taxonomies of Agent Proactivity, Persona, and Specialization

LLM-based MAS incorporate a rich taxonomy of agent behaviors and role archetypes to support division of labor and creative workflows [2505.21116].

- **Proactivity Classes**:  
  - *Reactive agents* remain in a wait-for-instructions mode, solely responding to external stimuli without initiative.  
  - *Proactive agents* continuously monitor contextual signals, taking initiative by proposing new tasks, suggesting changes, or alerting peers of opportunities.  
  - *Mixed agents* dynamically switch between reactive and proactive modes through policies or entropy-based thresholds.
- **Persona Archetypes**:  
  - *Expert* (high-precision domain specialist), *Creative Thinker* (divergent idea generator), *Critic* (evaluates feasibility or logic), *Coordinator* (mediates allocation and conflict).
- **Task Allocation & Creativity**:  
  - Coordinators allocate brainstorming to Creative Thinkers; Experts filter for correctness; Critics perform ranking or down-selection.  
  - Persona-driven pipelines enhance balance between novelty and feasibility in generative tasks.

Specialization and organic adaptation are observed through experience management: agents that excel at particular sub-processes see their role probability distributions updated, leading to emergent specialization and improved collective efficiency [2411.14033].

## 2. System Architectures and Communication Protocols

Architectural paradigms in LLM MAS include centralized, decentralized, layered, and hybrid designs, each with distinct trade-offs in scalability, robustness, and coordination overhead [2402.01680][2411.14033][2505.21116].

| Architecture           | Description                      | Strength    | Limitation             |
|------------------------|----------------------------------|-------------|------------------------|
| Centralized            | Orchestrator-agent manages state | Simplicity  | Single point of failure|
| Decentralized/Peer-Peer| Agents direct communication      | Robustness  | Consensus complexity   |
| Layered/Hierarchical   | Multi-level routing/validation   | Validation  | Latency, complexity    |
| Blackboard             | Public message board, dynamic selection | Context sharing | Scaling with roles    |

- **Message Passing**: Synchronous (RPC) and asynchronous (queue-based) methods support flexible coordination. Common message schemas include sender_id, receiver_id, type, payload, and timestamp [2505.21116].
- **Protocols**: Dynamic agent selection, majority voting, utility-based convergence, and auctioning are used for task allocation and consensus [2505.21116][2507.01701].
- **Example**: Task auctioning in decentralized settings, and blackboard-based architectures that achieve state-of-the-art reasoning performance with lower token costs than static pipelines [2507.01701].

## 3. Generation Techniques, Memory, and Planning Mechanisms

Creativity and collaborative effectiveness hinge on rigorous workflow design and sophisticated memory and planning modules [2505.21116][2504.01963].

- **Generation Techniques**:
  - *Divergent Exploration*: Agents maximize output novelty by broad sampling or prompt perturbation ($T$ increases in $p(w_t | w_{<t}) \propto \exp \frac{\log P_{\text{LLM}}(w_t | w_{<t})}{T}$).
  - *Iterative Refinement*: Multiple rounds of candidate output, scoring, and gradient-based updating of internal representations, e.g., $z^{(k+1)} = z^{(k)} + \eta \sum_j \nabla_z f_j(\text{decode}(z^{(k)}))$.
  - *Collaborative Synthesis*: Agents merge partial outputs, optimizing a global utility $U = \alpha N + \beta C + \gamma R$ (novelty, coherence, relevance).
- **Memory Systems**:
  - *Short-term*: Context windows for immediate history.
  - *Long-term*: Vector databases, retrieval-augmented logs, or symbolic stores.
  - *Shared Memory*: Centralized stores or blackboards for global context [2507.01701].
  - *Self-Controlled Memory*: LLM-based controllers decide what and when to store.
- **Planning Mechanisms**:
  - *Tree-of-Thought, ReAct frameworks*: Interleaved reasoning/action or explicit search trees [2504.01963].
  - *Parallelized Planning-Acting*: Dual-thread architectures with interruptible execution for real-time responsiveness [2503.03505].
  - *Graph-based and meta-learning policies*: Dependency graphs and meta-learned agent coordination [2503.10049].

## 4. Application Domains and Empirical Performance

LLM-based MAS are applied in problem solving, world simulation, data markets, control engineering, software engineering, and large-scale optimization [2402.01680][2511.16964][2511.13233][2505.19567][2504.04834].

- **Examples by Domain**:
  - *Software*: Multi-role development teams (e.g., MetaGPT), code review, and debug pipelines.
  - *Science*: Teams of researcher-assistants and critics for literature search or hypothesis generation.
  - *Economics*: LLM agents as buyers/sellers in data marketplaces, accurately reproducing real-world trading distributions [2511.13233].
  - *Engineering*: Supervisory MAS for controller design, with agents specializing in retrieval, reasoning, simulation, critique, and communication [2505.19567].
  - *Optimization*: Multi-agent systems achieve a 2.88× speedup in PyTorch inference tasks by pairing exploit-heavy search with error-fixing agents [2511.16964].
  - *Creative Generation*: LLM MAS surpass static and single-agent methods in knowledge, reasoning, and math, with reduced token usage [2507.01701][2505.21116].

| Domain        | MAS Architecture     | Metric/Task        | SOTA Example | Performance     |
|---------------|---------------------|--------------------|--------------|-----------------|
| Reasoning     | Blackboard          | MATH, GSM8K        | LbMAS        | 81.7% avg [2507.01701] |
| Software      | Layered/static      | HumanEval          | MetaGPT      | Competitive     |
| Data Markets  | Decentralized       | Trading metrics    | LLM-MAS      | Realistic distributions [2511.13233] |
| Optimization  | Specialized agents  | PyTorch speedup    | PIKE-B+EFA   | 2.88× [2511.16964]     |
| Engineering   | Hierarchical        | Control Design     | LLM-Agent-Controller | 83% completed [2505.19567] |

## 5. RL Optimization, Scalability, and Business Aspects

Reinforcement learning and graph-based policies support scalable, adaptive MAS; business incentives and privacy constraints shape deployment [2506.02718][2503.10049][2411.14033].

- **Multi-Agent RL (MARL)**:  
  - Critic-free algorithms such as MHGPO achieve higher stability and scalability over critic-based baselines (e.g., MAPPO), with group-based advantage estimation and flexible sampling (IS, FoF, RR) [2506.02718].
  - Joint policy optimization using reward propagation, agent-specific penalties, and meta-learning enhances coordination in both search and embodied domains [2503.10049].
- **Business/Privacy**:
  - MLAS protocol supports agent monetization via credit allocation (Shapley value), incentive formulas $I_e = \alpha\,\text{DataVal}_e + \beta\,\text{TrafficRev}_e + \gamma\,\text{IntelRev}_e$, and maintains privacy through local data stores, differential privacy, and SMPC [2411.14033].
  - Decentralized systems (e.g., AgentNet) combine RAG and evolutionary specialization in a DAG to enable privacy-preserving collaboration [2504.00587].

## 6. Evaluation Metrics, Security, and Risk Management

Robust evaluation frameworks and emerging threat models inform the reliability and safety of LLM-based MAS [2505.21116][2410.02506][2507.04724][2508.08127][2508.05687].

- **Creativity Metrics**:
  - *Novelty*: $1 - \max_{r\in\text{ref}} \frac{|x \cap r|}{|x \cup r|}$.
  - *Coherence*: Mean cosine similarity of adjacent sentence embeddings.
  - *Human-likeness*: Discriminator-predicted score $\mathrm{HL}(x)$.
  - *Distinct-n*: Fraction of unique $n$-grams.

- **General MAS Metrics**:
  - *Success Rate, Task Completion Time, Collaboration Efficiency, Resource Usage* [2402.01680].
  - *Accuracy, F1, Precision/Recall* in domain-specific tasks [2506.10467].

- **Security**:
  - *AgentPrune*: Spatial-temporal graph pruning reduces redundant/malicious communication, cuts costs up to 87%, and increases robustness under adversarial attack by 3.5–10.8% [2410.02506].
  - *BlindGuard*: Hierarchical, unsupervised anomaly detector for malicious agent propagation, effective without attack-specific labels [2508.08127].
  - *AgentXposed*: HEXACO trait modeling and behavioral interrogation for intention-hiding attack detection, outperforming MBTI/Big Five baselines by 8–15 pp in F1 [2507.04724].

- **Risk Analysis**:
  - Failure modes: cascading reliability, inter-agent communication failures, monoculture collapse, conformity bias, deficient theory of mind, and mixed-motive dynamics are formally defined and quantified [2508.05687].
  - Staged testing (simulation, sandbox, pilot, deployment) with convergent evidence is advocated for robust governance.
  - Key metrics: reliability cascade, communication fidelity, monoculture correlation, conformity error rate, and multi-agent risk scores.

## 7. Open Challenges and Future Directions

Outstanding research challenges include evaluation standardization, coordination protocol adaptivity, scaling communication, bias mitigation, theoretical guarantees, and collective intelligence synthesis [2505.21116][2402.01680][2411.14033][2508.08127][2508.05687].

- *Benchmark Unification*: Current lack of integrated benchmarks spanning text, image, and multi-modal creativity.
- *Bias and Ethics*: Insufficient mitigation of stereotype amplification; recommendation of embedded meta-personas for bias detection/mitigation.
- *Scalability*: $O(N^2)$ communication complexity requires topology adaptation and role assignment meta-learning.
- *Security*: Adaptive, topology-agnostic defenses needed for evolving adversarial threats; unsupervised and contrastive anomaly detection are promising directions.
- *Theory*: Formal limits of MAS synergy, triggers for emergent agency, and certification of “true” creativity remain open.

---
**References**  
- [2505.21116] Creativity in LLM-based Multi-Agent Systems: A Survey  
- [2402.01680] Large Language Model based Multi-Agents: A Survey of Progress and Challenges  
- [2507.01701] Exploring Advanced LLM Multi-Agent Systems Based on Blackboard Architecture  
- [2506.10467] Specification and Evaluation of Multi-Agent LLM Systems -- Prototype and Cybersecurity Applications  
- [2511.16964] Optimizing PyTorch Inference with LLM-Based Multi-Agent Systems  
- [2411.14033] LLM-based Multi-Agent Systems: Techniques and Business Perspectives  
- [2505.19567] LLM-Agent-Controller: A Universal Multi-Agent Large Language Model System as a Control Engineer  
- [2503.03505] Parallelized Planning-Acting for Efficient LLM-based Multi-Agent Systems  
- [2504.01963] LLMs Working in Harmony: A Survey on the Technological Aspects of Building Effective LLM-Based Multi Agent Systems  
- [2503.10049] Enhancing Multi-Agent Systems via Reinforcement Learning with LLM-based Planner and Graph-based Policy  
- [2511.13233] LLM-based Multi-Agent System for Simulating Strategic and Goal-Oriented Data Marketplaces  
- [2508.04652] LLM Collaboration With Multi-Agent Reinforcement Learning  
- [2506.02718] Heterogeneous Group-Based Reinforcement Learning for LLM-based Multi-Agent Systems  
- [2507.04724] Who's the Mole? Modeling and Detecting Intention-Hiding Malicious Agents in LLM-Based Multi-Agent Systems  
- [2508.08127] BlindGuard: Safeguarding LLM-based Multi-Agent Systems under Unknown Attacks  
- [2410.02506] Cut the Crap: An Economical Communication Pipeline for LLM-based Multi-Agent Systems  
- [2508.05687] Risk Analysis Techniques for Governed LLM-based Multi-Agent Systems  
- [2404.04834] LLM-Based Multi-Agent Systems for Software Engineering: Literature Review, Vision and the Road Ahead

Source: https://www.emergentmind.com/topics/llm-based-multi-agent-systems-c730e815-6eb0-4219-a652-47653955a70d