AITOPAI热报
ProductsOfficialSelected

Anthropic 如何构建多智能体研究系统

How we built our multi-agent research system Jun 13, 2025

Primary sourceAnthropic: Engineering
News· 11 sources
Read the original
Why we picked it

Anthropic 公开了多智能体系统的实战架构，做 AI 研究和智能体开发的团队可以直接借鉴其任务协调与分工设计，值得点开学习。

Anthropic 分享了其内部多智能体研究系统的构建经验。该系统利用多个 Claude 智能体协同工作，以加速 AI 安全研究。

Image · Anthropic: Engineering
Body

文章详细介绍了系统架构、智能体分工、任务协调机制以及在实际研究中的应用效果。这一系统展示了多智能体协作在复杂研究任务中的潜力，为 AI 研究自动化提供了新思路。

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Products" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Key entities and technical topics under continuous tracking include 多智能体, Anthropic, Claude, AI 安全.

Primary source verification: published directly by Anthropic: Engineering with first-party authority. Cross-source corroboration: 11 independent sources have covered this development (including Dario Amodei Blog, The Rundown AI, Ethan Mollick), confirming industry-wide signal strength.

#多智能体#Anthropic#Claude#AI 安全#研究自动化

Keep reading

Recent stories with the same tag or section
PapersBuilders13:57
Chinese contentarXiv 论文：任务委派式生成搜索效果不佳，直接工具调用更稳

做 multi-agent 系统的人应该看看这篇。它用 pre-/post-cutoff 数据集实测发现：委派别的模型干活有时还不如让同一个 LLM 直接调工具，还总结了查询被扭曲、信息传递丢失等失败模式，能帮你判断什么时候别乱委派。

#多智能体#RAG#LLM#工具调用
arXiv cs.AI@Tamara Stojanovska et al. (4)Source
PapersBuilders12:01
Chinese contentDecArena 基准发布：去中心化多智能体协作遇上 10 万工具规模测试

做智能体系统的人可以看看这个基准：它专门测没有全局编排器、只能靠邻居通信的场景，10 万工具规模下主流 LLM 智能体都翻车，RL 也救不回来。论文给的 GraphHarness 基线涨了 48.5 个点还省成本，思路对搭建大规模分布式智能体很有参考价值。

#多智能体#智能体#GraphHarness#去中心化协作
arXiv cs.AI@Jiali Cheng et al. (5)Source
PapersBuilders11:00
Chinese content把经济制度当作设计空间：arXiv 论文提出智能体经济组织的新坐标系

arXiv 上的理论论文，不是发模型那种。它把拍卖、市场这些制度拆成坐标维度，让你能设计出人类社会没有的经济结构。做多智能体系统、给 agent 设报酬和所有权规则的研究者会受益，做应用开发的可以先收藏。

#arXiv#多智能体#经济机制设计#智能体
arXiv cs.AI@Nikolos Gurney, Fred MorstatterSource
ProductsBuildersMulti-sourceSelected09:20
CanyonOS Scales Multi-Agent Workloads Per Agent, Not Per App

The non-obvious point is that agent-level autoscaling plus automatic cost tracing without code decorators removes two real operational pain points for LangChain and LangGraph teams, though adoption requires refactoring to one-file-per-agent. Developer and platform teams running multi-agent production workloads can start with the conversion assistant to migrate existing projects. The missing authentication and disabled dashboard login mean anyone self-hosting on AWS or Kubernetes must add access controls before exposing it.

#CanyonOS#agents#open-source models#OpenTelemetry
Event
shao__meng@shao__mengSource