Agents3h ago
Apple Research Finds Multi-Agent Teams Limit Experts
Unconstrained self-organizing LLM teams often fail to outperform single expert models. Apple researchers found that free interaction lacks the coordination needed for synergy. These systems frequently struggle with role confusion and redundant efforts. Practitioners should prioritize structured workflows over autonomous collaboration to ensure reliable performance in complex multi-agent deployments.