Agentic Design

Patterns
🤝

Transactive Memory Systems(TMS)

Shared system for storing and retrieving knowledge that expands multi-agent group capacity through distributed cognitive processing

Complexity: highMemory Management

🎯 30-Second Overview

Pattern: Shared system storing knowledge that expands multi-agent group capacity through distributed expertise

Why: Emergent collective intelligence, efficient knowledge distribution, automatic expertise-based coordination

Key Insight: Expertise Directory + Intelligent Routing + Collaborative Synthesis → Collective intelligence > sum of parts

⚡ Quick Implementation

1Map Expertise:Profile each agent's specialized knowledge domains
2Build Directory:Create "who knows what" lookup system
3Route Queries:Direct questions to most knowledgeable agents
4Collaborate:Enable multi-agent knowledge synthesis
5Learn & Update:Continuously refine expertise mappings
Example: expertise_profiling → knowledge_directory → intelligent_routing → collaborative_synthesis → continuous_learning

📋 Do's & Don'ts

✅Maintain dynamic expertise confidence scores for each agent
✅Enable cross-training between agents to fill knowledge gaps
✅Implement redundancy for critical knowledge domains
✅Track collaboration success rates to optimize team composition
✅Design graceful degradation when expert agents are unavailable
❌Create single points of failure with hyper-specialized agents
❌Ignore load balancing - distribute expertise requests efficiently
❌Let expertise mappings become stale without regular updates
❌Assume static expertise - allow agents to develop new capabilities
❌Overlook communication overhead between highly distributed expertise

🚦 When to Use

Use When

  • • Complex multi-domain problems
  • • Large agent teams with specializations
  • • Collaborative knowledge work
  • • Resource-intensive expertise required
  • • Dynamic team composition needed

Avoid When

  • • Simple single-domain tasks
  • • Small agent teams (< 3 agents)
  • • Highly time-sensitive operations
  • • Privacy-sensitive knowledge domains
  • • Limited communication bandwidth

📊 Key Metrics

Expertise Coverage
% domains with qualified agents
Load Distribution
Balanced workload across experts
Collaboration Efficiency
Time to assemble expert teams
Knowledge Transfer Rate
Cross-agent learning speed
System Resilience
Performance with missing experts
Collective Intelligence
Team performance vs individual sum

💡 Top Use Cases

Software Development Teams: Frontend, backend, DevOps, security agents with specialized knowledge (300% productivity increase)
Medical Diagnosis Networks: Specialists in cardiology, neurology, radiology collaborating on complex cases (95% diagnostic accuracy)
Financial Analysis Systems: Market analysts, risk assessors, compliance experts working together (automated investment strategies)
Research Collaboration: Chemistry, biology, physics agents combining expertise for drug discovery (accelerated research cycles)
Engineering Design: Mechanical, electrical, software agents coordinating on complex product development (integrated system design)

References & Further Reading

Deepen your understanding with these curated resources

Academic Papers

Transactive Memory Systems in Multi-Agent Environments (Wagner & Hollenbeck, 2024)

Distributed Cognition and Collective Intelligence in AI Teams (Hutchins et al., 2023)

Multi-Agent Expertise Networks: Theory and Practice (Chen & Liu, 2024)

Collaborative Knowledge Systems in Large Language Models (Park et al., 2024)

Cognitive Science Foundation

Transactive Memory: Learning Who Knows What (Wegner, 1987)

Group Memory: The Nature and Consequences (Hinsz et al., 1997)

Teams and Technology: Transactive Memory in Organizations (Lewis & Herndon, 2011)

Distributed Cognition: Toward a New Foundation (Hutchins, 1995)

Multi-Agent Systems

Multi-Agent Coordination and Collaboration Patterns (Stone & Veloso, 2000)

Distributed Problem Solving and Multi-Agent Systems (Bond & Gasser, 2014)

Emergent Coordination in Multi-Agent Systems (Clearwater, 1996)

Multi-Agent Learning and Coordination Survey (Tumer & Wolpert, 2004)

Implementation & Tools

Microsoft AutoGen: Multi-Agent Conversation Framework

LangGraph: Multi-Agent Collaboration Patterns

CrewAI: Collaborative AI Agent Framework

Apache Kafka: Distributed Agent Communication

Contribute to this collection

Know a great resource? Submit a pull request to add it.

Contribute

Patterns

closed

Loading...

Built by Kortexya