📄 中文摘要
在之前的研究中，作者介绍了一种每天早上5点自动运行的AI研究流程，利用Sonnet实现了高效的研究。然而，随着时间的推移，作者感到研究主题的选择有时显得肤浅，研究深度不够一致。为了提升研究质量，作者尝试了使用Opus作为编排器，组建一个由多个具有不同角色的智能体团队并行工作。结果显示，研究质量有所提升，但在成本和效率方面并未达到预期的投资回报率。
我尝试了 Opus 编排器并终结了它——多智能体系统的投资回报率
📄 English Summary
I Tried an Opus Orchestrator and Killed It — The ROI of Multi-Agent Systems
The previous research outlined an AI research pipeline that automatically runs every morning at 5 AM, utilizing Sonnet for efficient research. However, over time, the author felt that the topic selection sometimes lacked depth and consistency. To enhance research quality, the author experimented with using Opus as an orchestrator, forming a team of multiple agents with distinct roles working in parallel. The results indicated an improvement in research quality, but the cost and efficiency did not meet the expected return on investment.