---
格式版本: 2
标题: "A Multi-Agent Platform for Automated Enterprise Analytics and Insight Generation"
原文链接: "https://arxiv.org/abs/2608.18740"
发布日期: "2026-08-19"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Wed, 19 Aug 2026 09:49:58 UTC (154 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-20T15:35:24+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-20T15:33:55+08:00"
入库时间: "2026-08-20T07:35:24.844Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=latency&searchtype=all"
匹配关键词:
  - "latency"
  - "AI"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 28
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "论文主题为多智能体企业分析平台，与技术内容完全无关，仅因搜索命中latency而出现，未涉及超节点/AI Rack/供电/散热/互连等任何项目主题，应判无关。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-20T15:38:23+08:00"
AI主题相关性: 0
AI来源权威性: 10
AI新颖性: 5
AI技术细节: 5
AI商业部署信号: 0
AI完整性: 8
AI摘要: "论文提出一个基于CrewAI的多智能体企业分析平台，用五个专职AI代理依次处理自然语言查询、数据检索、分析和可视化洞察生成。在300个测试用例上，功能准确率为95.3%，较单智能体基线提升22.6个百分点，数据分析与报告聚合代理是质量关键。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-07T03:18:57.494Z"
采集批次: "2026年8月20日14点19分32秒"
采集批次ID: "20260820-141932-079"
去重键: "https://arxiv.org/abs/2608.18740"
---

## Computer Science > Artificial Intelligence

## Title:A Multi-Agent Platform for Automated Enterprise Analytics and Insight Generation

Authors:[Manoj N M](https://arxiv.org/search/cs?searchtype=author&query=M,+M+N), [Vijayakrishna S](https://arxiv.org/search/cs?searchtype=author&query=S,+V), [Manjunath Srinivas](https://arxiv.org/search/cs?searchtype=author&query=Manjunath), [Rohit Pahan](https://arxiv.org/search/cs?searchtype=author&query=Pahan,+R)

[View PDF](https://arxiv.org/pdf/2608.18740) [HTML (experimental)](https://arxiv.org/html/2608.18740v1)

> Abstract:This paper proposes a multi-agent framework built on CrewAI \[1\] for conversational business intelligence. Five specialized AI agents operate in a sequential pipeline to process natural language queries, retrieve and analyze data, generate visualizations via the Model Context Protocol (MCP) \[2\], and deliver actionable insights. The platform features a defense-in-depth security architecture for multi-tenant data isolation and a query parameterization mechanism for transforming conversational insights into reusable dashboard components. Evaluation across 300 end-to-end test cases spanning synthetic and production enterprise datasets demonstrates 95.3% functional accuracy, a mean response latency of 24 seconds, and a response quality score of 4.52/5.0 as assessed by an LLM-as-a-Judge framework, with a 93.0% hallucination-free rate, representing a 22.6 percentage point accuracy improvement and 20.2% quality gain over a single-agent baseline. Cross-model evaluation across four LLM backends and human expert validation confirm architectural generalizability and evaluator reliability. An ablation study confirms that the Data Analysis and Report Aggregation agents are the primary drivers of output quality.

| Subjects: | Artificial Intelligence (cs.AI) |
| --- | --- |
| Cite as: | [arXiv:2608.18740](https://arxiv.org/abs/2608.18740) \[cs.AI\] |
|  | (or [arXiv:2608.18740v1](https://arxiv.org/abs/2608.18740v1) \[cs.AI\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.18740](https://doi.org/10.48550/arXiv.2608.18740) |
| Related DOI: | [https://doi.org/10.1145/3841323.3841342](https://doi.org/10.1145/3841323.3841342) |

## Submission history

From: Manoj N M \[[view email](https://arxiv.org/show-email/4e685ee7/2608.18740)\]  
**\[v1\]** Wed, 19 Aug 2026 09:49:58 UTC (154 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.18740) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))