Lumina · 灵犀
个人 AI 秘书桌面应用
个人 AI 秘书桌面应用
每天要处理飞书消息、微信读书笔记、小红书收藏、邮件等大量碎片化信息，散落各处难以整合利用。需要一个真正「懂你」的AI助手，把所有个人信息源打通。 市面上的AI助手都是云端运行，隐私不可控，且无法连接个人数据源。我要做一个本地运行、数据不出机器、能接入个人全数据源的AI秘书。 统一管理所有个人信息源，AI基于历史数据做出个性化响应；MCP协议让能力无限扩展；持久记忆让AI越用越懂你，真正成为专属秘书。
Fragmented information across Feishu, WeRead, Xiaohongshu, email, and cloud drives is impossible to consolidate. You need an AI that truly 'knows you' — unifying all personal data sources. Existing AI assistants run in the cloud with no privacy control and can't connect to your data. I built a local-first AI secretary where data never leaves your machine, with real access to all your personal data sources. Unified personal data management with AI responses grounded in your history. MCP protocol enables infinite capability extension. Persistent memory means the AI gets better the more you use it.
Article Flow
自媒体文章全流程工作流管理平台
自媒体文章全流程工作流管理平台
自媒体创作者每天在选题、搜集、写作、去AI味、排版、配图、发布之间反复切换工具，流程断裂、效率极低，一篇文章从想法到发布常常耗时一整天。 现有写作工具只解决单点问题——能写不能发，能排版不能配图。没有一个工具能覆盖从选题到发布的完整链路，更没有AI辅助去AI味和多平台适配的能力。 10步流水线一站式管理创作全流程，让创作者专注内容本身；AI辅助写作+去AI味+多平台适配，单篇产出时间缩短60%以上；6大平台一键格式适配，告别手动调排版。
Content creators juggle disconnected tools across topic selection, research, writing, de-AI-ifying, formatting, images, and publishing. The broken workflow means one article can take an entire day from idea to publish. Existing tools solve isolated problems — you can write but not publish, format but not add images. Nothing covers the full pipeline from topic selection to publishing, let alone AI-assisted de-AI-ifying and multi-platform adaptation. 10-step pipeline manages the entire creative workflow in one place, letting creators focus on content. AI writing + de-AI-ifying + multi-platform adaptation cuts per-article time by 60%+. One-click formatting for 6 platforms — no more manual layout tweaking.
Stock Research
多智能体 A 股研究平台
Multi-agent A-share research platform
A股研究中，信息碎片化严重，研报、新闻、数据散落在各处，手动整合耗时且容易遗漏关键信息。需要一个智能系统自动收集、分析、生成研究报告。 将多智能体技术应用于金融研究领域，让不同的 AI Agent 各司其职：信息搜集、数据分析、报告撰写，最终生成专业级别的投资研究报告。 自动化 A 股研究流程，从信息收集到报告生成全流程智能化；多智能体协作确保分析维度的全面性；为投资决策提供数据驱动的深度洞察。
A-share research suffers from fragmented information — reports, news, and data are scattered everywhere. Manual integration is time-consuming and prone to missing key insights. An intelligent system is needed to automatically collect, analyze, and generate research reports. Applying multi-agent technology to financial research: different AI Agents handle information gathering, data analysis, and report writing, ultimately producing professional-grade investment research reports. Automates the entire A-share research workflow from information collection to report generation. Multi-agent collaboration ensures comprehensive analysis dimensions. Provides data-driven deep insights for investment decisions.
NoteAI
AI 驱动的个人知识库桌面应用
AI 驱动的个人知识库桌面应用
个人知识管理中，笔记越积越多却找不到、用不上。传统RAG每次都从原始文档重新检索和拼凑答案，没有积累，问了白问。 受「知识编译」思路启发：应让 AI 把零散资料沉淀为结构化知识库，而不是每次从原始文档临时检索。编译一次，持续增值。 三层知识架构让原始资料和最终答案之间有了「编译层」——AI主动维护的结构化知识库；AI问答基于编译后的知识而非原始文档，答案质量更高且越用越准；双向链接+关系图让知识真正连成网。
Personal notes pile up but become unfindable and unusable. Traditional RAG re-derives answers from raw documents every time — no accumulation, making repeated queries pointless. Inspired by a 'knowledge compilation' approach: AI should compile scattered material into a structured knowledge base, not rummage through raw documents each time. Compile once, appreciate forever. Three-layer knowledge architecture inserts a 'compilation layer' between raw materials and final answers — an AI-maintained structured knowledge base. AI Q&A leverages compiled knowledge for higher-quality answers that improve over time. Bidirectional links + relationship graphs turn knowledge into a real network.