How To Build an Efficient Personal Knowledge Management System

I believe that making an effort to establish a personal knowledge management system can be beneficial to everyone. Nowadays, we all face with too much fragmented information. We have to learn how to control the information by transforming quality content into knowledge and discarding momentary and useless information.

Every morning, when I wake up, open my mobile phone or laptop, enormous information rushes towards me. Originally, I feel so excited for I can instantly access the information on any topic from anywhere. It seems that I can obtain any knowledge in any field and solve any problem in the world.

I have read a lot of articles, posts and so on, and also saved a lot of them to my hard disk. Unfortunately, I gradually realized that although I spent plenty of time on various information, I did not become more knowledgeable. And sometimes, I even cannot remember that I have studied a topic despite the fact that I have stored many articles regarding the topic on my computer.

It is a sad truth that the information I have obtained just stored on the computer instead of saving in my brain and becoming my knowledge, which means that I still cannot freely recall it, organize it and use it whenever and wherever. Therefore I have to build a system that can help me to translate information into knowledge, and I did.

What is Knowledge Management

Knowledge Management is one of the hottest topics today in both the industry world and information research world. In our daily life, we deal with huge amount of data and information. Data and information are not knowledge until we know how to dig the value out of it. This is the reason we need knowledge management. Unfortunately, there’s no universal definition of knowledge management, just as there’s no agreement as to what constitutes knowledge in the first place.

There are over a hundred published definitions of knowledge management (Dalkir 2011) and perhaps the widely accepted one is developed by the Gartner Group and Bryant Duhon:

“Knowledge management is a discipline that promotes an integrated approach to identifying, capturing, evaluating, retrieving, and sharing all of an enterprise’s information assets. These assets may include databases, documents, policies, procedures, and previously un-captured expertise and experience in individual workers.” (Duhon, 1998)

Why We Need to Build a Personal Knowledge System

Although KM generally refers to achieving organizational objectives or business result, I mainly focus on personal KM in this article to help us develop ourselves more and gain more knowledge.

After building my KM system, I keep improving and modifying it, which makes me become more efficient and help me completely solve the following issues:

  1. Amass information without creation.

    As I have mentioned before, every day I read a lot of articles or posts from websites, social media, and Google search. When I found an excellent article, I will promptly save it to Evernote, a digital notebook I have used before, for further reference. After many years of accumulation, there are thousands of notes in my Evernote notebook. However, I did not read most of them for the second time after saving them.

  2. No output, no sharing.

    For example, when I was dealing with a problem, I always referred to multiple resources and even did certain experiments to conclude the final solution. After the task had been done, I did not organize all the information regarding the solution and write down how I found it. Also, I did not share it with someone else. After a time, I will forget all about these: the references, the experiments, and the solution. All my efforts do not achieve their maximum value.

The Key To A Good KMS


The above picture is a DIKW (abbreviation of Data, Information, Knowledge, Wisdom) pyramid from David McCandless, which illustrates the hierarchy of data, information, knowledge and wisdom. It is important to note that there is a definite boundary in the middle of the pyramid: Data and information are what we collected, Knowledge and wisdom are what we created.

The essential part of a good KMS is output. Reading books, watching videos and attending lectures are all activities for acquiring information. Even though books and lectures belong to wisdom or knowledge in the DIKW model, they are the authors’ wisdom and lecturers’ knowledge. They are not ours. The best way to transform the information into our knowledge is to interpret something new to us based on what we have learned and share them.

Interpreting helps us organize and digest information. Therefore it can translate information into our embedded knowledge. And sharing is not only a contribution to the information age but also our motivation to keep outputting better works, which is the reason why I believe that information becomes knowledge via organizing and sharing.

Working Flow

Basic Flow


The above figure illustrates the primary working flow of my KMS, which is “Collect, Organize, and Output.” The three phases correspond to the DIKW model. After the organization, the information that we collected transforms into our knowledge. Then if we applied the knowledge in practice, such as writing an article or resolving a problem, it would eventually become our wisdom.

The figure also shows the iOS apps I currently use in each phase. Although any app would be changed in the future if I found a better one, they are currently the best choice for me after testing many similar apps.


DEVONthink is the most important application of this system. It is the central storage and the perfect Document Library / Note Taking solution. All information that is processed by me is classified and saved in DEVONthink.

Evernote acts as a container for Business Cards, a Read Later client, and the temporary storage for raw materials, especially for Web Clips. This is because Ervernote’s Web Clip feature still is the best one in the market. Materials which I do not study them and write notes on them are called “raw material,” I only keep them for further reference.


We face two types of information: Active information and passive information. Active information is what we intend to acquire, such as googling a keyword. By contrast, passive information is pushed to us by its’ senders, such as twitters, posts and so on.

In the collection phase, the primary task is to judge which information should be kept for further reading. And it is also important that we should not spend too much time browsing the passive information.

Every day, I spend about thirty minutes browsing rapidly the articles from LinkedIn, Facebook, Twitter and Feedly. If I found an interested one, I would clip it to DEVONthink or Evernote for later reading.

I digitize paper documents via Scanner Pro and save them to iCloud, so that the document can automatically be imported to DEVONthink.

After this phase is finished, all the information which needs to be processed in the next phase is stored in my inbox, which is composed of DEVONthink, Evernote and Things.

Processing and Organization


It is important to establish a theoretical system before planning our knowledge management system because it will guide us on how to define the tags and how to organize notes, etc. Otherwise, we would get lost in the forest because the only thing we could see is trees. For example, one time I felt that my notebooks are too disordered to organize all my documents well. So I googled on how to classify notebooks and had been convinced by an article, I organized my notebooks according to this article. Unfortunately, after a few weeks, my notebooks evolved into another chaos. This is not because of the article’s fault, it is because I only learned ‘HOW’ from this article and did not know ‘WHY’. Establishing a theoretical system will let us clear the reason why we should do this and why we should do that.

When I was in primary school, my teacher told me a method on how to read a book: Reading a book should be firstly “from thin to thick” and then “from thick to thin”, which comes from a famous Chinese mathematician named Hua Luogeng.

“From thin to thick” means that learning a book should also extend its content by complementing and annotating. For example, if we cannot understand a term or a concept, we could google it or study it from related books, and then write down our understanding. Thus, the book becomes thicker.

After a period of study, we have understood every word, every sentence and every chapter. But this does not mean that we apprehend the book. Because:

Finally, there is a process from “thick to thin”. The things that have already been learned must be chewed, digested, organized, re-examined, integrated, refined, and critically identified. Compare to what you have learned in the past and find out what new content and new methods have been added. After this, we will find that the book seems to become thin from thick. — Hua Luogeng

“From thin to thick and then from thick to thin” becomes the bedrock of my whole knowledge management system. It is not only suitable for reading books but also applicable for technical research and fragmented reading.


For each of the collected material, we need to decide whether we should study it or just keep it for further reference. The decision depends on the subjects on which we are currently focusing. For example, if I am currently concentrating on web development, an article named “10 tips to build a fast HTML 5 app” is definitely worth reading carefully. And “Healthy dinner recipes”, I am not interested in cooking, but these recipes seem perfect and maybe someday I will need to prepare a feast, so it would be better to save it in my “Reference” notebook.

Organization mainly includes classification and complement. Besides highlighting and noting down the important points in an article, it is also necessary to expand on certain contents or words to help us understand more deeply. If a post contains the term “DES”, for instance, which I do not know its meaning, I will google it and note down the explanation.

After processing a material, I should save it to an appropriate location and tag it, in other words, classify it into the proper category. In DEVONthink, I am using two levels for classification: Groups and Tags. Regarding how to build a personal note system, this is another topic that is worth writing the other article to describe it in detail. Simply put, the principle of my note-taking solution is that the notes should be obtained or consulted again easily.

The most important step is to review the organized materials periodically. Without review, our hard disk would become a landfill instead of a treasure trove of knowledge.

There are not too many apps needed in this phase. Following are the apps I frequently use:

  • DEVONthink
  • MarginNote Pro
  • XMind Zen
  • Workflowy

Besides these Apps, I also use SONY DPT RP1 as my active reading device. This is a fantastic digital paper device and your eyes will eventually thank you for choosing it.

After the processing and organization phase, all the organized and processed information is saved in DEVONthink, MarginNote and Workflowy.


As I have mentioned before, the output is an essential stage of knowledge management. Output, such as writing an article, telling a story or delivering a speech, compels us to structure and reorganize the knowledge that we have learned, and to depict it in an explicit way that others could easily understand. After this stage, we grasp the knowledge because we have known it, understood it and used it.

I use Typora for most of my writing tasks. Typora is a decent Markdown editor app which can make me concentrate on typing and enjoy writing. I used Ulysses before and I still love Ulysses’s free Sheet feature. However, the high price and the lack of support for Tables and Graphics moved me away from it.


Collection, organization and output constitute a whole system of knowledge management. A lack of any one of these three parts will cause us to get half the result with double efforts in learning.

Found this post useful? Please share this post because it will help more people find it. Thanks!

I’m Axton, an entrepreneur and a content creator. I believe in the power of delivering value to you, the reader. If you like this sharing please share it with one of your friends. You can also find me on Twitter and on my Chinese Blog

Read more

Jina Reader API 的四种用法 | 「智图派」

Jina Reader API 的四种用法 | 「智图派」

构建知识库,或者分析各种文章数据,是大家使用 AI 很重要的一个应用场景,因此我们常常会需要用到爬虫去爬取某个网站上的内容,现在,Jina 推出了一款非常简单好用的获取网页内容的工具,你只要把网址给它,它就能把网页内容整理成很适合大语言模型使用的格式,简直是构建知识库的利器。今天我就给大家介绍四种用法。分别包括在自动化工作流中使用以及在 AI 智能体中使用。 Jina Reader API 的网址是 读取器 API,你可以直接输入你需要爬取内容的网址 URL 在这边进行测试,输入之后直接点击按钮「获取内容」,就可以在右边得到结果了。 而他的用法也很简单,就是把你需要爬取内容的网页的 URL 写在 的后面就可以了。 什么是 Reader API Reader 是将任何URL转换为LLM友好的输入,只需简单添加前缀 无需付费即可获得改进后的适用于智能体或 RAG 系统的输出。

lock-1 By Axton
Coze 工作流获取 Tweets 详解 | 「智图派」

Coze 工作流获取 Tweets 详解 | 「智图派」

Coze是一个可以不用代码来搭建AI聊天机器人的平台。我在这期视频当中已经详细地讲解了Coze的各种功能,以及如何用到GPT-4 Turbo的模型。感兴趣的朋友可以回看一下。 自从 Twitter 被马斯克买了之后,如果你想从 Twitter 网站的外面来获取推文信息,也就是用 Twitter API,就需要每个月充值100美元了。但是 Coze 却很大方地让你可以免费地用部分 Twitter API 的功能了。今天我就先给大家演示一个可以批量获取多个 Twitter 作者的推文的工作流,把这个工作流嵌入到你的 AI 聊天机器人里面。比如我这里加了两个 Twitter 账号,我的和马斯克的,这样就可以一次性地获取到我们两个人的推文了。 今天我首先给大家演示一下在 Coze 上可以用到的 Twitter API 的功能,然后再演示一下批量获取推文的工作流,最后我们再聊一聊 Coze 的局限性。 Plugins 的基本功能 getUsers 首先第一个 "get users"

lock-1 By Axton
建立微信 AI 工作流 | 支持 10000 个文件,Assistants API v2 发布 | AI 快讯20240418

建立微信 AI 工作流 | 支持 10000 个文件,Assistants API v2 发布 | AI 快讯20240418

点击进入「AI 精英学院」,做一个会 AI 的人- 新视频发布:微信如何通过 Make 与 GPT、Claude、知识库无缝集成 - 支持 10000 个文件,Assistants API v2 发布 - 我们需要为生物技术的“ChatGPT时刻”做好准备 - NVIDIA推出基于安培架构的两款新GPU,RTX A400和RTX A1000 - Google地图将利用AI帮助您找到隐蔽的电动车充电器 微信如何通过 Make 与 GPT、Claude、知识库无缝集成 没想到海外的自动化平台居然也支持微信公众号。Make 工作流自动化平台,不但支持微信,还支持的挺完善,确实有些让人意外。本期视频演示如何使用 Make 获取微信公众号信息、与 OpenAI 知识库联动、

By Axton
DALL.E 推出图像编辑以及风格推荐功能 | Sora 再现竞争对手 | AI 快讯20240404

DALL.E 推出图像编辑以及风格推荐功能 | Sora 再现竞争对手 | AI 快讯20240404

点击进入「AI 精英学院」,做一个会 AI 的人- DALL.E 推出图像编辑以及风格推荐功能 - “多次尝试破解”:实验室揭示AI安全特性如何被轻松绕过 - 前Snap AI负责人推出Higgsfield,挑战OpenAI的Sora - MIT研究人员开发新计算技术以简化蛋白质工程化过程 使用DALL·E编辑图片 原文链接:Editing your images with DALL·E | OpenAI Help Center OpenAI 的推文包含两项功能的视频演示: You can also get inspiration on styles when creating images in the DALL·E GPT.

By Axton