Amazon researchers evaluated four agent architectures across 13 standardized enterprise workflows. The single agent achieved a 67.4% task success rate, while adding high-quality skills raised it to 73.1% and adding low-quality skills reduced it to 59.8%. The best multi-agent architecture reached 75.6%, but the more coordinated configurations consumed up to 4.7 times as many tokens and increased latency by 2.6 to 5.8 times. No architecture delivered a statistically significant improvement across all 13 workflows, leading the researchers to recommend starting with a single agent and adding complexity only when it produces a measurable benefit.
Source: Read the original article
Disclaimer: this content is for information purposes only and is not financial advice. Cryptocurrencies are highly volatile: you may lose all of your capital. Always do your own research. Legal notice
