Maggie Appleton 写设计、人类学和智能体界面时，习惯先画清楚隐喻，再谈工具。把「Tending Evergreen Notes in Roam Research」整理成可阅读的中文笔记：问题在哪、界面默认会把人带去哪、落地时该改什么。原站导航、评论回响和广告已去掉。

A tour through the Evergreens

Active users of Roam Research interested in learning new techniques and workflows that will help them connect and organise their ideas

Just like a good seventeenth-century English garden, this Digital Garden has two sides to it – the private and the public. Everything I post here first sprouts in my Roam Research garden. If you have never heard of Roam, first stop by Anne-Laure LeCunff’s articles on Roam for Metacognition , and the more hands-on The Beginner’s Guide to Roam . Otherwise, the rest of this post will be completely bewildering. While many digital gardeners choose to make their

Anatomy of an Evergreen Note

Roam is where my unfiltered and unshaped thinking happens. It acts as a place I can externalise my thoughts to see what 一个稳妥判断是, clarify those thoughts, and connect them together.

There are many names for systems like this; second brains , tools for thought , or personal knowledge management practices. Perhaps “intentional notetaking” sufficiently captures it.

Working in a Constantly Evolving System

My approach is heavily influenced by the Zettelkasten system; a method that focuses on constructing a web of declarative, connected notes that act as building blocks for creative output. If you are new to the idea of Zettelkasten, I strongly recommend you read the introductory material . Understanding this approach to knowledge work is far more profound and transformative than any app will ever be. Sonke Ahren’s How to Take Smart Notes book, as well as And

This TLDR is you focus on writing atomic , declarative notes in your own words . The titles of these “ Evergreen ” or “ Zettel ” notes are statements; things you believe based on what you’ve read and experienced in the world. Things like “ Naming is the hardest part of programming ”, or “ Fitness culture has replaced religion in modern society ”. You follow this up with a paragraph or two explaining the statement and any literary references that led you to

Page Prefixes

You then combine these Evergreen notes into sequences. Each note can belong to multiple lines of thought. Connecting them together easily allows you to create outlines for articles, papers, or other forms of knowledge work. By using this core collection of beliefs as a starting point to build arguments and logical conclusions, you’re never faced with a blank page. Realisations emerge from the bottom up.

There’s plenty of hype and chatter about Zettelkasten systems in the Roam community. However, I haven’t seen many walkthroughs of well-developed databases filled with Zettels. Two notable exceptions to this are Joel Chan and Beau Haan , both of whom have very thoughtfully mapped the qualities of a Zettelkasten into a Roam system.

Keyboard Maestro

Most other tutorials lack practical examples outside the realm of productivity and note-taking. They fail to show what a collection of real Zettels looks like, how they emerge from research, and how they connect to lead to meaningful output.

I’m sharing my system here to put one more tangible example out into the world. I’ve been working with Zettels in Roam for over a year and have 150+ of them written Which is still a meagre beginner pile compared to the 90,000 Niklas Luhmann wrote over the course of his career. #ZettelGoals . They have helped me clarify my beliefs, synthesise my understanding of other people’s work, see connections across many different texts, spot holes in my arguments, an

1. Capturing Links

My system is not an ideal or universal template. It certainly has flaws I’m oblivious to and will learn over time. If you notice potential flaws, I would love to hear your thoughts only if you’ve been working in a Zettelkasten for 6+ months and creating public-facing work with it. It may not fit what you want and need from a system. But taken with a grain of salt, this post may give you ideas for your own Roam-based Zettelkasten.

The easiest way to show you this system is to walk you through it. I’ve recorded a quick video tour here, but you can also find the written version below.

值得单独记下的观察

• ⌆ for speculative outlines of original creations
• ➽ for original creations themselves – drafts that are/will be public at some point. Essays, digital garden notes , conference talks, video tutorials, and livestream/podcast outlines all fall under this umbrella

落地时建议先做的 5 件事

1. 先写清这个工具在强化思考还是在替人思考。
2. 默认交互不要只会谄媚：该追问、该给反例、该要求证据。
3. 智能体界面要露出推理步骤，而不是只给一个光滑答案。
4. 知识发布用可生长的笔记，而不是一次性营销长文。
5. 改产品前先画隐喻：用户以为自己在做什么，系统实际在做什么。

和智能体产品怎么接

龙虾PRO做 OpenClaw 落地时，最该从这类笔记里拿走的是「别把聊天框当唯一界面」。数字员工要能追问、能验证、能把过程摊开，而不是只负责说好话。

本文侧重全链路风控方法论。落地时请用自身业务单据做回放验证，不要把示例阈值直接当生产策略。 相关：风控体检 · 方案资源