Original Summary

CRM IMAGE Check My CRM Obsidian Over the years, I accumulated more than 3,600 LinkedIn connections. But recently, I realized that I could no longer answer some very basic questions: When did we connect? Where did we meet? Why did we connect? What did we talk about? The number of connections kept growing, while the context behind those relationships slowly disappeared. So instead of trying to add more people, I started building a system to better remember the people I had already met. I took part of the network data I had organized from LinkedIn and connected it to my existing Obsidian-based LLM Wiki. That became the first version of what I am calling a personal “Relationship OS.” What it does Each person has their own page containing more than just a name, company, and job title. I can connect information such as: How and where we met What we talked about Their interests and areas of expertise Related companies, events, topics, and projects When we last communicated Notes that may help me reconnect meaningfully later The people pages are also connected to the other information already stored in my Wiki, including projects, meetings, events, companies, and ideas. The attached image shows the current relationship graph inside my Obsidian vault. Searching relationships by context The most useful part is being able to search my network based on fragments of memory rather than names. For example, I can ask: Who was the PM I spoke with at that AI event? Who in my network has experience in manufacturing? Who is connected to a specific company, project, or topic? Who have I not spoken with recently, but might have a meaningful reason to reconnect with? The LLM searches the connected notes and suggests people who may match that context. Instead of manually browsing thousands of profiles or old message threads, I can describe what I remember about the relationship. A few principles I do not want to compromise While building the prototype, I established three important rules. First, importing the source data again must never overwrite notes I wrote manually. Second, people should not be deleted simply because we have not spoken for a long time. An inactive relationship is still part of my personal history. Third, I deliberately excluded automated outreach. I do not want to build another tool that automatically generates messages, schedules follow-ups, and turns every relationship into a sales pipeline. The system can help me remember and find people. But I should still decide whether, when, and how to contact them. The product should support the relationship—not automate the human part of it. Where I want to take it next This is still a personal prototype rather than a public product. My next step is to connect more context, including project records and conversations I have had with LLMs. Eventually, I want the system to do more than locate a person. I want it to surface relationships that may be relevant to something I am currently researching, building,

  • 情报分类：工作与职业机会
  • 分类依据：内容涉及招聘、求职或职业发展
  • 信息来源：Reddit · SideProject
  • 发布时间：2026/9/20 17:26:20