Author’s Note:
This article itself is a practical application of the workflow described within it.
The core ideas were structured using the Tana #WritingFunction framework mentioned in the text, which was used to organize viewpoints, examples, and the main body of the article. The logic was then validated and the wording refined through discussions with Claude.
The central ideas, personal experiences, usage cases, and opinions in this article are all original. Any AI-assisted content has been manually reviewed and confirmed.
Before We Begin
“Should I choose A or B?” This is probably a question that comes up in almost every discussion about tools in any specialized field. In the world of note-taking and knowledge management, you constantly see similar debates: Logseq or Obsidian? Notion or Roam Research? And for me, the A and B that kept me torn for a long time were Heptabase and Tana.
To be honest, this was probably a kind of obsession. We always hope to find an all-in-one tool that can solve everything at once—capturing, organizing, thinking, and producing output, all in a single place. In fact, many tools really are evolving in this direction.
But the problem is that when you try to make one tool cover every scenario, you end up with something that is only “okay” at each of them, and truly excellent at none. This isn’t the tool’s fault; it’s a contradiction in our own expectations. More importantly, we’re easily led astray by features. We focus too much on “what this tool can do” and forget the more fundamental question: “What do I actually want to achieve?”
In the end, when it comes to knowledge management, techniques and features are only the surface. What really matters is how you build your way of thinking, how you understand knowledge, and what you can ultimately distill into something that truly belongs to you.
Both Heptabase and Tana are products I genuinely love. They are platforms I want to use for a long time to learn and to accumulate knowledge. But these two products are completely different, both in philosophy and in functionality. How should I choose? This question troubled me for a long time and consumed a great deal of my time and energy.
I remember repeatedly trying to make one of them my all-in-one solution, and repeatedly failing. I kept searching for “the perfect fit,” only to realize that each had irreplaceable strengths. Switching back and forth left me anxious and inefficient.
Yet it was precisely this period of hesitation and struggle that gradually helped me figure out a few things.
Why do I need them? What problems do I actually want them to help me solve?
As the answers to these questions became clearer, the dilemma that had bothered me for so long suddenly dissolved. I stopped obsessing over “which one to choose” and instead decided to use both. That was the original starting point of this article.
Today, I won’t go into detailed introductions of the specific features of either tool—those can easily be found on their official websites and in tutorials. What I really want to talk about is how I understand these two products from the perspective of product philosophy and real needs, and why I ultimately chose to “practice both.”
If you’re struggling with a similar question, perhaps this article can offer you a different way of thinking.
Preface
In 2023, I left my company. All kinds of notes I had accumulated over more than ten years were returned along with the computer. That moment actually felt pretty good, because I no longer had to maintain all that scattered, messy information. What followed was a chance to start from scratch—for me, that meant no historical baggage and no need to worry about data migration. I could rebuild a knowledge system of my own, centered around the fields and questions I truly care about, and reconstruct my thinking framework from the ground up.
Not long after, I came across Heptabase and Tana, and began trying to use a single tool to handle all of my knowledge management. Over the past few years, I’ve switched back and forth between them, hoping to find the most suitable approach. But every attempt lasted only a few months at most. I always felt something was off, yet I couldn’t quite articulate what.
So where exactly was the problem?
It took me a long time to gradually realize that the issue wasn’t the tools themselves at all. These two tools follow completely different philosophies. One emphasizes space, the other structure. They’re not really comparable in the first place. Yet I had trapped myself in the obsession with “all in one,” always wanting a single tool to cover every scenario. The result was that nothing was properly addressed, and nothing was done well.
To be honest, this shift in mindset took me quite a long time.
That’s why in this article, I want to revisit the topic of knowledge management. I want to talk about my years of going back and forth between tools, my confusion along the way, and how I now think about the question of “how to choose a knowledge management tool.”
Heptabase vs. Tana
The characteristics of these two products are very distinct, and their differences are substantial.
Heptabase is essentially a tool for spatial thinking. It completely frees notes from the logic of linear text, allowing us to organize information on a two-dimensional plane. It’s a bit like breaking down complex problems, placing all the variables, relationships, and logic within the same field of view.
For example, the kind of case analysis scenes we often see in movies and TV shows:
Ideas that were originally scattered across different documents are turned into movable cards. You can reposition them at any time, create connections, and form groups. This process is, in fact, a visualization of your thinking path.
But what ultimately convinced me to pay wasn’t this interaction style itself—it was the product’s underlying understanding of “knowledge.” The core of Heptabase is not “recording information,” but “understanding knowledge.” It treats thinking as a spatial activity that requires seeing the whole picture, discovering connections, and building structures. This is completely different from the traditional note-taking logic of “write it down and you’re done.”
The founder, Alan, explains this very clearly in his article My Vision Project Meta: the ultimate goal of note-taking tools should not be storage, but helping people understand complex things. I strongly resonate with this idea.
By contrast, Tana takes a completely different path. What it focuses on is not the “big picture,” but “structure”—or more precisely, how to organize information through structure.
I’ve always preferred outline-based tools, for a simple reason: they’re fast and flexible, ideal for capturing things in the moment. During my years at the company, I used Logseq and OmniOutliner for a long time. They worked fine for note-taking, but had one fatal flaw—information was flat and fragmented, lacking semantic connections. You could record a lot, but everything remained isolated nodes. You knew they were there, but it was hard to form a systematic knowledge network.
Tana’s core innovation lies in Supertags, and this was also the main reason I decided to pay for it. Supertags turn each node from just a piece of text into an “object” that can carry structure and attributes. This means we can move information from the level of simple “recording” to the level of “modeling.”
For example, when recording a book: in traditional outline tools, you can only write the title plus some notes. In Tana, however, you can define the author, publication year, domain classification, reading status, and even link it to your own thoughts and writing. This isn’t just a functional difference—it’s a cognitive upgrade. We begin to understand knowledge in a structured way, instead of merely piling up information.
Of course, on the other hand, Tana is also the tool with the steepest learning curve I’ve ever encountered. It requires a certain level of abstract thinking and modeling ability. You need to define your own ontology, design structured fields, and build query logic. This process is far from easy and inevitably involves repeated trial and error, patience, and persistence.
To be honest, I gave up on Tana several times halfway through. It wasn’t until the past year, as my understanding of structured thinking deepened, that I finally started to use it smoothly.
Going Further
There are many tool products out there, but the ones that truly endure are never those with the longest feature lists. They last because they embody a unique way of thinking. At their core, every good tool is an externalization of how its founding team understands a particular domain. How they see the problem determines what the product looks like—and also how you, as a user, are guided to think when using it.
In other words, a tool is not just a passive container. It also expresses a way of thinking.
Heptabase and Tana are two classic examples. One spatializes “thinking,” the other structures “knowledge.” You might say, isn’t that just a difference in interface and interaction?
No. What lies beneath is two completely different views on knowledge management and information processing. Heptabase focuses on “seeing relationships.” It believes that thinking requires a global view, that hidden connections must be discovered in space. Tana, on the other hand, focuses on “defining relationships.” It believes knowledge must be structured and that information should be organized through explicit semantics.
One is bottom-up emergence; the other is top-down construction. Ultimately, both tools are answering the same question:
How should we handle information so that it truly becomes knowledge?
Heptabase: Spatializing Thinking
As mentioned earlier, what Heptabase does is liberate notes from linear text. That may sound trivial, but in fact it is a redefinition of what “thinking” is.
In Heptabase, every piece of information is a card. Cards can be freely placed, grouped, and connected on a whiteboard. As the number of cards grows, you begin to notice something: you are no longer merely recording things—you are building a map of your thinking.
Why do this? Because the human brain does not process complex problems in a linear way. We need to see the whole picture, discover hidden connections, and move back and forth between different pieces of information before new understanding can emerge. Traditional notes are page after page of documents; your field of view is limited. Heptabase externalizes this process.
Here’s an example. Recently, I was organizing a whiteboard about Japan’s interest rate policy, triggered by news that the Bank of Japan was considering raising rates. I started wondering: what is the logic behind this? What chain reactions might it cause?
So I began placing cards on the board: the background of this rate hike, its triggers, how a stronger yen might affect international markets, several major interest rate policy shifts in Japan’s history… Each new card raised new questions. Gradually, a structure began to emerge on the board. I realized that I had developed a completely different understanding of “the lost thirty years” and “the ten-year cycle.” Bits of information that I had once read in isolation suddenly connected into a coherent line.
This is the core value of Heptabase. It doesn’t help you store notes; it helps you clarify your thinking. When you turn fragmented information into movable cards and continuously adjust their positions and connections on a whiteboard, you are really doing one thing: turning the thinking process in your head into something visible and manipulable.
This is something traditional note-taking tools cannot do. In those tools, once you’ve written something down, it’s basically finished. In Heptabase, writing is only the beginning. The real thinking happens when you reorganize the cards. You discover new relationships, raise new questions, overturn old conclusions. Thinking no longer stays inside your head—it becomes a structure you can actually see.
That’s why Heptabase is easy to start with, but hard to master. What’s hard about it? You have to actively build. You have to be willing to spend time breaking problems apart, repeatedly adjusting the positions of cards, and wrestling with your own thoughts in the process. It’s slow and tiring.
But that is what deep thinking really looks like.
By the way, the whiteboard example I mentioned is only a starting point. You can also check out the official “Chip War” case study—it’s more complete and better demonstrates the power of this way of thinking.
Tana: Structuring Thinking
If Heptabase helps you see your thinking, then Tana helps you organize it.
So what is its core idea? It turns the logic of your thinking into structures that can be defined and reused.
In Tana, every piece of information is a node. But unlike traditional outliners, each node here can carry semantic structure through Supertags. You can define what this node is, what attributes it has, and how it relates to other nodes. When these nodes reference, link to, and nest within one another, you gradually see them weaving into a dynamic knowledge network.
This means your notes are no longer dusty text that sits unused after being written, but structured information that can be organized, reasoned over, and even trigger actions.
Let me share a few of my own examples.
Daily capture of thoughts
I use the tag #Signal to record my daily thoughts and summaries. But I don’t just record them — I also add structured annotations:
- Domain: Which field does this thought belong to? For example, macroeconomics, product design, AI applications…
- Context: In what situation did this idea arise?
- Content: What exactly is the thought?
Take the note about the Bank of Japan possibly raising interest rates as an example. I don’t just record the fact that “the BOJ may raise rates,” but also tag its domain (macroeconomics), its context (pressure from yen appreciation), and the specific analytical content.
The benefit of this approach is that later I can retrieve and organize these thoughts by domain, by context, or by time — from different dimensions. Each piece of information is no longer isolated, but a knowledge node with an “identity tag.”
Solidifying a writing framework
Writing is an even more typical case — for me, it is essentially a process of structured thinking. So I created a dedicated #WritingFunction tag, breaking article writing into a set of fixed questions:
Core questions (to answer every time):
- Core message: What is this article trying to say?
- Background & context: Why write it, and in what situation?
- Key insight: What new discovery do I have?
- Core viewpoint: What is my position?
Optional modules (used as needed):
- Supporting evidence: What materials support my argument?
- Analogies: Can analogies help understanding?
- Perspective shift: Would it be clearer from another angle?
- Cognitive progression: What further thinking can this article provoke?
This framework is essentially my “thinking framework” for writing. It ensures that each piece is not improvised on the spot, written wherever it goes, but instead becomes a systematic reasoning process. I no longer have to start from scratch every time to figure out how to write; I just fill in, expand, and refine within the framework. This very article was drafted under this Writing Function and then refined through discussions with AI.
The framework itself is also iterative. When I find certain questions repeatedly useful, I solidify them into the template. When some questions prove unhelpful, I remove them.
Modeling company analysis
Using the same logic, I also built a model for company analysis — an ontology structure for fundamental analysis. What does that mean? It means I defined the dimensions needed to understand a company.
For example, in the Palantir case shown above:
- Business model: What are the main products? Where does revenue come from?
- Customer base: Who does it serve — governments, enterprises, or individuals?
- Growth path: How does it expand — through product iteration or market expansion?
Once this model is built, I can quickly analyze any new company with it. Instead of rethinking “what should I look at” every time, I simply fill in and compare according to the template. This goes beyond traditional note-taking and turns my way of thinking into a reusable structure.
So what is Tana’s true value in these scenarios? I believe it is definitely not about recording information, but about making your thinking logic explicit and structured — turning it into something that can be built, reused, and iterated. In other words, these frameworks composed of Supertags in Tana are the externalization of how we think.
This is also why Tana has such a steep learning curve. It requires strong structured thinking: you need to know how you think before you can build that logic using Supertags, fields, and queries. This process takes time. I personally gave up several times along the way, and only in the past year did I truly start to use it smoothly.
But honestly, the process itself is very valuable. I gained a much clearer understanding of how I think. It constantly forces you to ask: how do I actually understand a problem? What is my thinking framework? That alone is already extremely worthwhile.
Back to the Fundamentals: What Is the Logic of Knowledge Management?
After talking about these two products, we actually need to return to a more fundamental question:
How do we truly understand “knowledge”? What do we think we are managing, and how should it be managed?
This question is important. We cannot decide what form tools take, but we can decide our own philosophy. A person capable of independent thinking should have their own view of knowledge. That understanding is the real logic behind choosing tools.
Many times, you’ll find that tools recommended by others look extremely tempting to try. But once you actually start using them, they feel awkward no matter what you do.
Where is the problem? It’s not really the tool itself, but the fact that it doesn’t fit the way you think. Or rather, you may not yet have a sufficiently clear methodology to make full use of it. That’s not a bad thing. On the contrary, it’s a reminder that you should pause and think about your own logic of knowledge management.
We don’t have to chase every new tool or keep experimenting endlessly. We should first clarify how we ourselves think.
Essentially, whether it’s knowledge management or building an understanding of the world, the first step is probably not to aim for a specific tool. It is to find your own logic and philosophy. Only then can you truly choose—or even shape—tools that match your way of thinking.
For example, if you want your knowledge management process to move from chaos to clarity, you may need to first lay everything out and look for connections in space—then Heptabase might be what you need. But if your thinking style is more standardized, where you define a framework first and then fill it in, Tana might suit you better.
Neither approach is right or wrong. What matters is that you know which one you want.
Don’t Be Obsessed with “All in One”
Knowledge management itself is a huge topic. It consists of many different scenarios, and everyone emphasizes different aspects. When you stack them together, the logic becomes very complex.
For me, there are only two core concerns:
- How to quickly capture and structure information to form my foundational knowledge material, preparing for future personal “model training”;
- How to integrate this material around a specific question or domain into a complete knowledge framework, enabling deeper research into a topic.
These don’t sound special. Tana and Heptabase both seem capable of doing them. But in practice, neither alone works well. Heptabase’s whiteboards are powerful for organizing information and divergent thinking, but inefficient for quick capture and structured processing. Tana excels at recording and structuring, but because its product design is node-centric, it is not good at presenting a holistic view.
Trying to focus on details while also maintaining a bird’s-eye view within a single tool is, at least for now, unrealistic—or more precisely, very awkward to use. So this year I changed my approach and started using Tana and Heptabase in parallel. Tana handles early-stage quick capture and structuring as a foundational database; Heptabase handles later-stage synthesis on whiteboards, building a global perspective around specific questions and domains.
Interestingly, after running this setup for a while, I realized that the copying and pasting between two tools didn’t add as much workload as I had feared (compared to forcing everything into an all-in-one solution).
Why? Because in Tana, every type of information has its own structured fields. The act of recording is already a process of understanding and digesting information. Through this “Q&A-style” approach, I end up with highly complete material. When importing it into Heptabase, very little adjustment is needed—it’s immediately usable.
This brings us back to the core point mentioned earlier: tools are merely carriers of logic. What really matters is how you understand knowledge and how you build your thinking framework. Once that’s clear, Tana and Heptabase are no longer competitors, but complementary tools for different scenarios.
That’s why I say: for knowledge management, don’t cling to the idea of all-in-one. Compared to the small cost of moving information between multiple tools, the cost of all-in-one solutions is often much higher. They not only reduce efficiency, but also easily confine your thinking within the boundaries of the tool.
More importantly, when you’re obsessed with finding the perfect tool, you’re actually avoiding a deeper question. Tools are always just tools. Your understanding of your own thinking is what determines how much value a tool can truly deliver.
So before choosing any particular tool, I suggest first clarifying your needs and your thinking process, then finding the tools that best fit each stage. Even if that means using two or three tools, as long as together they support your thinking workflow, that’s a good choice.
Compared to a mature system, the cost of one or two extra tools is trivial.
In Closing
From a product perspective, all note-taking tools ultimately boil down to “create, delete, update, query” plus “views and presentation.” There won’t be huge differences in functionality. But that’s not the point. The real question is: why do we do knowledge management in the first place? It sounds abstract, but once you figure it out, many troubles simply disappear.
If your goal is to store more information, any tool will do. But if your goal is to think better, then tool selection becomes a different question: can it make my thinking process explicit? Can it let me see how I think?
That is what truly attracts me to Heptabase and Tana.
They are not just helping me manage information—they are forcing me to understand my own way of thinking. Heptabase lets me see the spatial structure of thought; Tana lets me define the logical framework of thought. Once the system integrates with how you think, you no longer agonize over which feature to use, how to categorize things, or where to put them.
So, back to the two protagonists of today’s article: if you ask me which one to choose, Heptabase or Tana, my answer would be—don’t rush to choose. Spend some time first understanding how you think and what kind of cognitive support you need. Then you’ll find that whether the answer is A, B, or both A and B, you already have it.
And finally, one more thing: the endpoint of knowledge management is not building a perfect system, but becoming a clearer thinker. Tools are only the starting point. Thinking is the destination.
