GuShiio.com Wen| Wang Zhiyuan
My understanding of knowledge bases is divided into five types:
Open knowledge bases, personal knowledge bases, small organization knowledge bases, departmental knowledge bases and enterprise knowledge bases.They have different categories and different application scenarios.
01
Let’s talk about opening the knowledge base first. Some people pick a theme, organize a large amount of content, and then share it publicly. This is a public knowledge base. It has many uses, the most common one is to gain users.
For example:
When DeepSeek R1 first became popular, someone sorted out the founder’s background, how the company developed, how to use the products, and local deployment tutorials, and put them out for everyone to see. Behind this, they might want to promote their own products or services.
How to push? When users learn how to deploy DeepSeek R1, relevant tools or services are readily recommended in the knowledge base. If users find it convenient, they may become potential customers.
Or, they use these content to attract target users, and eventually sell more products or add several service subscriptions.
Another use is to increase website traffic and SEO.
Some companies publish valuable content to attract crawlers from search engines (such as Baidu and Google) and optimize website rankings. With more links, it is easier to be discovered.
In addition, someone uses it to collect user ideas.When users read articles in the knowledge base, new problems may arise; when others see these problems, they can use them to improve their products; for them, the knowledge base is an interactive place where they can receive user feedback and promote product upgrades.
However, the quality of content in such libraries varies. Sometimes you may even encounter data fraud, so you need to be careful and carefully identify it.
In short, opening a free knowledge base saves the cost of traditional content marketing, not only provides disorderly to orderly services, but also promotes the company’s products, which can be said to kill two birds with one stone.
02
A personal knowledge base, also known as a private knowledge base, is your own “knowledge treasure chest” that is used to store the knowledge you have learned, the useful information you see, and the experience you have accumulated in daily life. This definition is important.
I currently observe two trends:The first trend: collect more, organize more, but output less.
When building a personal knowledge base, many people are keen to collect various materials and include them in software such as cubox, ima, notion, and flying book, but very few people actually summarize and summarize the articles they write into their own opinions.
Collecting behavior is gradually widening the gap between adults in the use of knowledge bases. Most content that seems useful is actually not very helpful. They just create a sense of gain.
If you don’t believe me, you can stop and look through your knowledge base. Is it full of undigested content? These contents are like rotten pork, tasteless to eat but a pity to discard.
The second trend emphasizes focusing on collection and focusing on application.
Such people selectively collect information, quickly break it down into atomic units, and apply it to actual scenarios, making the knowledge base a tool to improve efficiency and capabilities, rather than a pure information warehouse.
Usually, if you think of a good idea today and a new idea tomorrow, you record it immediately and combine it with your personal experience to create an exclusive knowledge base.
In this process, the knowledge base is like a melting pot, gathering fragmented ideas into towers and eventually forming articles to support tasks and projects, forming a virtuous cycle. I have observed that this approach implies two understandings:
- atomized information
- Seize the gaps in knowledge
The concept of atom-ize originated from ancient Greece. In 1803, John Dalton proposed that the world is composed of particles, which form all things through different combinations. I often use this method in my work and daily habits. It is like establishing a formula for positive feedback. It has low resistance, quick start, and a strong feel for doing things.
For example, when writing, some people like to finish it in one go, but I write a little every day, write it down on a small card, and rest when I am tired. This is an atomization approach.
On Bilibili, I saw someone breaking other people’s content into information cards, reassembling it, and expressing it in their own style. This not only absorbs knowledge, but also creates value. This is also an atomic method.
“Seizing the gaps in knowledge is the process of making information grow gradually.
Ideas are often scattered and easily forgotten when left in note-taking software. I am used to sharing ideas into the community, collecting different opinions, and adding them to the content, so that the fragmented content will gradually become a whole.
An example:
A few days ago, I saw Descartes ‘famous saying and I thought, therefore I am. At first, I thought that I was referring to the individual in everyday language.
Later, after reading “The Great Conversation”, I discovered that the “I” that Descartes referred to was actually a pure and unquestionable consciousness, a thinking thing. This in-depth understanding made me re-examine my previous notes.
Another example is brand marketing.
In the past, people often came to me to talk about brand marketing, but they left after the chat, leaving nothing behind. Later, I began to sort out the key content, ranging from a hundred words to a few hundred words, thinking that it might be useful in the future.
At first, there were only scattered records and I didn’t pay much attention. Looking back now, these notes have formed a system; however, the system is not complete yet, and something is missing in some places. I will try to find a way to feed back the problems to gain inspiration.
This is the role of atomization and knowledge gaps: fragmented ideas gradually accumulate, fill in gaps through sharing and communication, and finally form a complete system. Knowledge can be pieced together like a jigsaw puzzle.
Let’s talk about the small organization knowledge base.
Some AI tools on the market today, such as Perplexity, Zhihu Direct Answer, Direct Access, or Ima.copilot and other search products, all have one thing in common: they provide a personal space. This space not only allows you to collect your favorite knowledge points, but also create your own themes and even share them with others.
If we extend this function a step further, it can become a knowledge base used by a small team of 3 to 5 people. They can add content and organize data in it. I call it a group-based knowledge base.
This form is already quite common, such as: for example, someone specifically organizes marketing-related knowledge, or various other topics. It is different from a large public knowledge base, in which content only circulates in a small scale.
A few days ago, in a community I joined, they used Mini programs to build a knowledge base. If you want to view the content, you need to enter a password, and the password is only open to community members. I think this form will become more and more popular in the future, because it is easy to manage, can also be closely integrated with the community, taking into account content organization and communication.
03
Some parts of the departmental knowledge base seem to be the easiest, but they are also the most difficult.
The easy part is that you can quickly convene a meeting of department members and easily collect a large amount of explicit knowledge, such as documents, PPTs, etc. These content is not difficult to obtain, but most of them can only be archived and have limited practical use.
In the early years, I stepped through this pit when building a departmental knowledge base. Everyone threw all the messy information in, and as a result, the knowledge base became a warehouse that no one cared about. Later, I adjusted my thinking and set a principle when establishing standards:Resolutely problem-oriented 。
This means that the content of the department’s knowledge base must come from problems encountered in actual work, rather than being preset out of thin air or filled in to make up the numbers.Based on this idea, I divided the content of the departmental knowledge base into two categories:
One is standard rule documents, such as department processes, specification descriptions, etc., which are convenient for new colleagues to quickly start using or for collaboration by other departments; the other is a question index, which is similar to the department’s exclusive Hundred Questions Encyclopedia. Anyone who encounters doubts at work can find answers or clues here.
Building a departmental knowledge base is the first step, and management is equally challenging.
How often is it updated? When will it be updated? What to update? These issues are not just as simple as arranging the responsible person, but also require clear rules to support them. Otherwise, the knowledge base can easily become a pile of dead files.
Later, I discovered that the key to making the departmental knowledge base truly work lies in everyone’s participation。For example: regularly collect questions, encourage each other to ask questions, and even organize brainstorms to sort out problems encountered at work and gradually enrich the content.
In this way, the knowledge base can become a living tool that everyone is willing to use.
After understanding the logic of the department’s knowledge base, you can also understand the source of the AI+ knowledge base.In fact, the essence of AI+ knowledge base is to replace cumbersome and repetitive questioning work and improve efficiency by automating common questions.
Smart customer service is a common example in daily life.
When shopping on Taobao and Jingdong, you asked: How to deliver the goods? When is the delivery date? AI will directly retrieve answers from the knowledge base and respond quickly, saving time and effort.
The same is true for departmental knowledge bases. Newcomers ask how to use processes and tools, and AI can automatically give standard answers, so old employees do not have to answer them repeatedly. This is why we often say that building a departmental knowledge base is crucial.
Nowadays, the content of the precipitation department is not as complicated as before. I have noticed that various office tools such as nails provide in-terminal search services for individuals and enterprises, and users can directly ask questions to AI assistants.
This is an important step in the office tool revolution because it makes information acquisition more efficient and smarter.Perhaps in the future, more products will launch similar functions of intra-terminal search, which will further change the way we work.
04
Many people believe that the enterprise knowledge base is the integration of the knowledge bases of multiple departments.
For example, some are public, such as vernacular, to store help documents and promote communication; some are used internally, and are filled with business processes, technical materials, customer information, etc.
I don’t see it that way. An enterprise is essentially a product container, and its core revolves around the product.
To be precise, the enterprise knowledge base should take the product as the main line, connecting relevant business processes, technical materials, and customer feedback. To put it bluntly, it is more like a user manual.
An example:
You use a company’s product. When you encounter after-sales problems, in the past, you might need to make phone calls, transfer labor, or even find the sales person. Now that you have the corporate knowledge base, you can directly open it and enter questions like flipping through a user manual., you can find the standard answer in a few seconds.
So, it is a product-centric, user-oriented utility.
However, I have seen many companies taking detours and turning the corporate knowledge base into an internal system, allowing employees to ask questions about the knowledge base. This in itself is no problem. After all, improving internal efficiency is also an important requirement.
However, the problem is, you must not confuse the internal and external knowledge bases and carry out them at the same time; otherwise, it may ultimately lead to the fact that it is not convenient for employees to use and cannot effectively serve customers. Both sides want to grasp it, but ultimately fail.
Regarding the future development of enterprise knowledge bases, I think it will move towards a platform-based + knowledge base +Agent solution. At present, there are many enterprise-level development platforms, such as:
- TARS-RPA-Agent (Automation Anywhere’s RPA+AI agent platform)
- CubeAgent (communication technology platform, including automation functions)
- Torq (Safe Ultra-Automation Platform)
- Lanma Technology’s AskXbot (focusing on knowledge base management and AI agents)
These platforms are characterized by deep integration of platform development, knowledge base storage and AI agent technology, and are specially designed to optimize the internal efficiency of the enterprise.
In contrast, tools like Byte Button and Smart Spectrum GLMS are more oriented towards consumer applications or general AI tools, lacking focus and integration capabilities for enterprise knowledge base management.
However, recently I found that like buttons, they have also begun to try workflow-related functions.This makes me think that some common tools may also evolve in the direction of platform + knowledge base + workflow + agent in the future.
But at present, their positioning and depth are still not as clear as enterprise-level platforms.
But no matter what, the enterprise knowledge base is an indispensable part of many companies in the future; due to the popularization of AI and agents, the role of the knowledge base in it is becoming increasingly important.
So how to build a knowledge base? What kind of knowledge base should you build? Who does the knowledge base serve? How do you want to use it?I hope these thoughts can provide you with some new ideas.
