金融工程 2026-03-24 AI 投研应用系列之五： OpenClaw 投研场景下微信 ClawBot 与飞书接入对比 相关研究报告 《AI 投研应用系列之四：OpenClaw 投研实践——从部署到应用》—— 2026.03.19 证券分析师：刘晓锋 电话：13401163428 E-Mail：liuxf@tpyzq.com 分析师登记编号：S1190522090001 证券分析师：马自妍 电话：010-88321787 E-Mail：mazy@tpyzq.com 分析师登记编号：S1190519070001 报告摘要 微信 ClawBot 与飞书均可作为 OpenClaw 的通信接入方式，为投 研人员提供与 AI 智能体交互的通道。本文从投研应用视角出发，对两 种接入方式的配置方法、推送与问答能力、功能局限进行对比，并结 合实际场景提出选型建议。 推送场景对比 微信 ClawBot 支持本地记录长期保存、多模态消息与定时任务， 但存在 24 小时交互时效限制，且 Markdown 渲染不完整。飞书支持群 聊推送、可读写云文档与多维表格，并支持完整 Markdown 渲染。 问答场景对比 微信 ClawB ...

2026-03-19 金融工程 AI 投研应用系列之四： OpenClaw 投研实践——从部署到应用 相关研究报告 报告摘要 OpenClaw 作为开源 AI 智能体框架，正在从对话工具向执行工具 演进。本文从投研应用角度出发，介绍了 OpenClaw 的部署方案、数 据源接入以及投研应用场景实践。 OpenClaw 部署与使用 对比纯本地、WSL2+云端模型、纯云端三类方案的适用场景与优 劣势，并以 WSL2+云端模型为例，完整演示环境搭建、OpenClaw 安 装、大模型配置及飞书集成流程。 介绍了 Tushare、AkShare 等数据 Skill 及其他金融数据库的配置 方法。 OpenClaw 投研应用 持仓监控报告推送：实现自动化数据采集、分析与报告推送。 量化策略回测与优化：通过行业动量+拥挤度轮动策略测试 OpenClaw 的量化回测与参数优化能力。从测试体验来看，当前大模型 自动生成的脚本仍可能存在逻辑错误或数据处理偏差，需人工反复调 试与修正，仍需进一步提升。 前沿因子挖掘：构建因子发现 Agent，实现论文自动抓取与潜在 因子思路提炼，总结因子逻辑、构建方法与评估可实现性。随着大模 型能 ...

Core Viewpoint - The article discusses the application of OpenClaw in investment research and office automation, highlighting its capabilities in financial data integration, stock selection, document management, and report generation through various skills and automation processes [1][2][3]. Group 1: OpenClaw Applications - OpenClaw enables multi-platform deployment for investment research, including local Windows, Mac, and cloud servers, facilitating financial data access and analysis [1]. - The system automates the extraction of unstructured data from PDFs, complex data cleaning in Excel, and the generation of structured reports in Word, significantly reducing manual errors and time costs [2][7]. - OpenClaw's intelligent agent framework automates the entire process from data collection to structured output, enhancing the efficiency of investment research [3]. Group 2: Automation of Office Document Processing - The article details how OpenClaw can automate workflows involving docx, pdf, pptx, xlsx, and canvas-design skills, addressing the challenges of cross-software collaboration and data transfer [6][7]. - Each skill serves specific functions: - docx for creating and editing Word documents, enhancing collaboration [8]. - pdf for extracting and processing data from PDF files, reducing manual data entry [8]. - pptx for generating and editing PowerPoint presentations, streamlining report presentations [8]. - xlsx for managing and analyzing Excel spreadsheets, improving data accuracy [8]. - canvas-design for creating high-quality visual designs, supporting various design needs [9]. Group 3: Practical Examples - An example illustrates how OpenClaw can read a PDF financial report, summarize the content, generate visual charts, and compile everything into a Word document, showcasing its end-to-end automation capabilities [26][33]. - The process involves extracting key information from complex documents and transforming it into structured outputs, demonstrating the tool's effectiveness in handling large volumes of data [33].