How To Build Your Zettelkasten to Master AI
“Now, here, you see, it takes all the running you can do, to keep in the same place. If you want to get somewhere else, you must run at least twice as fast as that!” - The Red Queen
Those who don’t learn to use AI, won’t be able to keep pace with those who do. AI writing is rattling the writer’s scene. It might be one of the strongest disruptors for the reason that it will be able to automate almost everything that isn’t truly original.
I strongly hold the belief that the same is true for the Zettelkasten Method: Those who don’t build an integrated thinking environment, won’t be able to keep pace with those who do.
I stumbled over a statement that perfectly express how AI and the Zettelkasten Method work together. The first one is:
You can’t automate what you can’t articulate.
I couldn’t put it in better words. Take a short look at the prompt shown in this video (should start at 53:10):
How did Nicolas Cole develop this highly intricate prompt? By obsessing over this topic for years and years on end.
Does it already dawn on you what a possible role a Zettelkasten could have had in developing the prompt? The Zettelkasten would provide the integrated thinking environment during the learning and building phase that made this prompt possible.
Think of AI as an intricate calculator. Your role is to master formulas, logical frameworks, inputs, and outputs and how, when and why you can or should apply them. Without mastering your mental tools, the calculator becomes useless, or worse, it may inflate your self-perceived competence, leading to errors. You can’t automate what you can’t articulate.
How could Cole have used the Zettelkasten Method to work on this prompt? This is the simple, yet powerful setup:
- The central note is the prompt that he wants to develop. This is the tool that he wants to build, or, as I once called it, A little machine.
- From this central note, he links to the background reasoning that justifies and explains each element of the prompt. These are atomic notes.
- On separate structure notes, he would develop the knowledge required in each domain. Domains include: The ghost writing landscape, prompt theory, marketing theory and related fields.
There is no directional workflow, since each direction is possible and should be exploited. There are several use cases:
- First, Cole works on his prompt, sparking ideas related to domains like ghostwriting practices, prompt engineering, and marketing theory. While iterating, he creates atomic notes for background reasoning, which he integrates into structure notes for each domain.
- Alternatively, he refines an atomic note’s background reasoning, uncovering connections to other ideas. By linking atomic notes, he enriches his knowledge network, capturing insights for future reference.
- Finally, he is hooked by a foundational book on a specific domain such as marketing theory. Each step of processing this book offers opportunities to improve the prompt and enrich background reasoning.
This structure wouldn’t just be an awesome foundation for his learning process. Set up properly, he could transform this into staff training for his company, multiplying the effect of his learning.
Why Zettelkasten.de Cannot be Automated
This is why you won’t find any AI content on this page. The only use case for AI on this page is to serve me as an editor. The ideas presented here are primary research, based on more than 15 years of work, many of which I spent obsessed with my research. I worked all day every day on driving my research forward and developing the tools for it.
The actual domain of my research is to learn about the nature of the good life and transform this knowledge into tools to improve the way of living on three levels:
- What is a good action? (e.g. a particular training method)
- What is the habit that integrates the action into your way of living and how to build it? (e.g. build morning routine, build a grocery shopping routine, etc.)
- What are the individual beliefs that can provide the foundation for a healthy way of living? (e.g. Beliefs like ‘I sin by binge eating and must punish myself through exercise’ versus ‘I work through training and fuel myself with eating.’)
To enable my work, I must build thinking tools and an environment that integrates knowledge across disciplines.
In dire need of thinking tools, I never found a comprehensive system for working with knowledge. Obviously, there are domain-specific tools, which I must learn and master adequately. However, I don’t think this is enough. Moreover, I think it is possible to learn a general understanding of knowledge, similar to an athlete who does general fitness training in addition to his sport-specific training.
I am confident in this statement because of my coaching experience. I am teaching people the Zettelkasten Method who use it for all various disciplines: Hindu philosophy, theology, fiction writing, Christian counselling, general fitness training, personal self-education, gardening, chemistry, biology and more. As long as I can understand the words, I can access the knowledge layer. This is necessary to teach the method beyond superficial mechanics, which provide little value anyway.1
This is what you find on this page: Primary research on the nature of knowledge and how to master tools to deal with knowledge.
An example is the concept and the method of soft modelling. Hard modelling is what you do, if you are an engineer or mathematician. These models involve number-crunching and adhere to strict truth criteria. Soft modelling on the other hand is about mapping qualitative relationships of elements of reality. Since they are similar to metaphors, they are confused with metaphors. Roughly, models are tools for rational thought, metaphors are tools to convey an emotional relationship.
One insight about soft models is that some models enable model-specific claims about completeness. Example: If you map something on a circle, you will see covered angles and missing gaps. If you explored the full 360 degrees, you can make the model-specific claim that you explored everything. Obviously, you haven’t explored every possible model. Therefore, each completeness claim is specific to the model used.
A use case for the above from my work: I reverse engineered the upper body basics presented by Ido Portal in a deliberately cryptical series of videos. (Example 1, Example 2) The exercise selection is guided by covering every shoulder angle with both straight and bent arms.
Obviously, I don’t know if I am 100% right. But I know with 100% certainty that I didn’t miss anything on the sagittal plane relating to the shoulders.
Zettelkasten.de cannot be automated by AI, because it won’t come up with evidence layering or organising your reading with your zettelkasten in mind.2
What’s the Verdict Then?
I initially aimed to use Nicolas Cole’s video to explore AI’s automation of Zettelkasten tasks, like finding novel connections. However, writing this article led me to another interesting take on the relationship of AI and the Zettelkasten Method.
This article has two parts:
- A plain description on how to use your Zettelkasten to work on AI prompts.
- A reason why AI can’t write anything for this page.
I combined both parts deliberately. AI has a specific relationship to our zettelkastens. I am seeing a pattern, but can’t put my finger on it yet.
I can’t write a comprehensive article about the relationship like I did about the Second Brain and the Zettelkasten Method. So, I am very much looking forward to feedback and your opinion.
-
“Ghost in the box? Spectators visit. They get to see everything, and nothing but that - like in a porn movie. And the disappointment is correspondingly high.” (note 9/8,3) ↩
-
I asked Grok “I have a zettelkasten. How should I adapt the way I read and why?” The answer is predictably disappointing: https://grok.com/share/bGVnYWN5_d97d39a7-0ad6-46a7-9774-68b6ff48e1d7 ↩