Multi-Agent Aggregator Workflow
Demonstrates multi-agent collaboration using the Aggregator agent to synthesize insights from specialized AI agents.
Quick Start
cd examples/aggregator-workflow
export OPENAI_API_KEY="your-key"
go run main.go
Overview
This example showcases a Research Synthesis System with:
- 6 specialized expert agents - Technical, Data Science, Business, Security, Ethics, Domain experts
- 3 aggregation strategies - Consensus, Semantic, and Weighted synthesis
- Conflict resolution - LLM-powered reasoning to resolve expert disagreements
- Semantic clustering - Group related insights thematically
- Final synthesis - Comprehensive analysis with recommendations
Comprehensive Guide: See Classifier & Aggregator Examples for strategy selection, configuration options, and advanced patterns.
Aggregation Strategies
| Strategy |
When to Use |
Key Feature |
| Consensus |
Building agreement |
Identifies common ground, resolves conflicts |
| Semantic |
Understanding themes |
Groups by similarity, preserves relationships |
| Weighted |
Expert prioritization |
Applies expertise-based weights |
Files
main.go - Complete multi-agent workflow
config.yaml - Expert definitions, aggregation settings
research_synthesis_output.json - Detailed results (generated)
Example Output
=== CONSENSUS AGGREGATION ===
Consensus Level: 0.83
Content: Based on expert consensus, LLMs are transforming software
development with 83% agreement on key impacts...
Conflicts Resolved: 2
=== FINAL SYNTHESIS ===
Key Insights:
1. Technical Impact (Confidence: 0.92) - 40-60% acceleration
2. Security Considerations (Priority: 0.95) - New vulnerability classes
3. Business Implications (Confidence: 0.75) - Positive ROI at scale