A YC Startup, Khoj[1], recently did a ShowHN[2]. It works and has several options, including an Obsidian Plugin and for Emacs. Unfortunately, it drags my computer, and I can't do anything else much once I run the local server. Perhaps, once I upgrade out of my Intel iMac, I will return to it.
(I'm not affiliated to Khoj.)
Does this sentence imply that some data of my vault will be sent to OpenAI? [nut sure my company will be fond of that]
tossing one's private notes to openai seems odd for such a tool that is marketed this way:
> Obsidian is the private and flexible writing app that adapts to the way you think.
> Obsidian stores notes on your device, so you can access them quickly, even offline. No one else can read them, not even us.
I have no affiliation to Foam. It's just awesome.
Yeah... I don't think we need more self congratulating note taking questions /s
Given that my colleague already had a wealth of experience using Obsidian and I made some contributions to LangChain, a very nice open-source Python framework for LLM communication, we decided to unite forces - I always enjoyed creating rich user interfaces while my colleague, Patrik, dealt with the part of the project that my UI needed to render. This is how our work on ODIN began, but what is Obsidian, how did a god from Norse mythology get in here, and why is TypeScript great? Let's find out.
Although I didn’t have such extensive experience working with Obsidian, I fell in love with it at first sight - a pragmatic and user-friendly interface, nice colors, a wide and very open open-source community and other goodies - what else does a good note-taking application need? With these thoughts in mind, I began to study how Obsidian works from the inside, how it can be expanded, and how it can be connected to other technologies - such as LangChain, in our case.
I was very pleasantly surprised when I discovered a well-made Developer API that allows you to create your custom community plugins that can easily connect to the application itself. I was even more pleasantly surprised when I discovered that all this is web-based and has support for React, which I worked with, and Svelte for its fans. Although the official documentation is not too full of details and sometimes you can get lost in it, a very convenient UI styling system and strict TypeScript type definitions, coupled with IntelliSense, which I always use, is a game-changer.
What's notable is the ease of creating plugins in real time; npm scripts allow you to instantly detect code changes, and the Obsidian Hot Reload plugin speeds up development several times. In addition, Obsidian provides developer tools like popular browsers, simplifying the debugging process by accessing the console and directly inspecting elements on the screen. This flexibility in the Obsidian API allows developers to bring their creativity to life, unlocking the full potential of Obsidian to create unique note-taking experiences and more. Honestly, it feels just like making a web application in a modern and robust environment. I'll miss this when I get back to regular web development.
That’s all great, but still, why ODIN? Honestly, it’s quite simple — we liked the project naming style that has existed in Memgraph for a long time — for example, MAGE, also known as Memgraph Advanced Graph Extension, seemed to us a damn original name, sharp as volcanic glass — and so, ODIN was born, from now onwards known as Obsidian Driven Information Network, an Obsidian plugin for knowledge management. This name became the real starting point of our project, representing the goal that we were trying to comprehend and the idea that we were trying to realize.
In our R&D process, while working on ODIN, we figured out that data storage is the core of the app's architecture, without which neither the backend nor the frontend would make any sense. Even though we thought a lot about traditional databases such as PostgreSQL, the Obsidian knowledge graph was the main inspiration to try Memgraph, a graph database that excels in modeling complex relationships and connections, making it the ideal choice for creating and navigating intricate information networks.
Even I, an ordinary student with a passion for developing user interfaces, was able to play around with both Memgraph and the query language called Cypher. I highly recommend trying the visualization that Memgraph Lab provides, even if you are far from both database graphs and software development in general.
By seamless integration with LangChain, we’ve tried to leverage Memgraph's speed and efficiency to transform standard notes into interconnected knowledge graphs empowered by the flexibility of embedding search. This enables you not only to traverse your notes effortlessly but also to extract valuable insights by interacting with LLMs through LangChain implemented in the backend using Python and FastAPI. With Memgraph as our data storage solution, ODIN redefines the way we store and harness knowledge, offering a platform for building an intelligent and dynamic personal knowledge management system just from your notes. Isn’t it wonderful?
ODIN introduces a multitude of functionalities to improve your knowledge management experience. At its core, ODIN provides a chat-like prompt bar for answering questions about the data stored in your knowledge graph as if you were just chatting with your knowledge collection.
With ODIN's advanced graph visualization capabilities provided by Cytoscape, an amazing library for graph visualization with complete TypeScript compatibility, you gain a whole new view of your Obsidian vault, effortlessly navigating through the complex web of your notes. This extends to individual files view, where you can delve deep into specific topics from your markdown notes. Furthermore, ODIN expands the Obsidian native drop-down menu by adding new advanced functions. Link prediction, as its name suggests, generates connections to other markdown files in your entire vault, enriching your note's content with thematically relevant references. Moreover, you can extract thought-provoking questions from your markdown file, fostering deeper contemplation. Lastly, the node suggestion capability enhances your note-taking by highlighting thematically connected nodes, forging meaningful connections between highlighted text and related knowledge nodes, all while promoting a more comprehensive understanding of your stored information. On the other hand, clicking the node from the file view will highlight related sentences directly in your opened note. These features perfectly represent the goals we were trying to achieve and have proven to be useful in our daily lives, which means they might be of interest to other developers as well as anyone else interested in organizing their knowledge in a better way.
The Obsidian community (including you, reading this article!) is the lifeblood of the platform's evolution and continued success. At the forefront of this community-driven ecosystem are the invaluable community plugins, just like ODIN. These user-generated extensions enhance Obsidian's functionality, addressing specific user needs and preferences. Becoming a part of this thriving community is not only about reaping the benefits of these plugins but also contributing to its growth. By actively participating in discussions, providing feedback, or even crafting your own plugins, you can play a pivotal role in shaping the future of Obsidian, enriching the collective knowledge management experience, and ensuring that this vibrant community continues to flourish.
The history of the creation of this project was long (and sometimes painful to remember), but if you were to ask me if I would do it again, I would without any hesitation. Creating plugin-type interfaces is a breath of fresh air in frontend development, which still fits perfectly not only into open-source contribution but also into professional experience along with career growth. Although this project could be done better in many ways, it will remain a reliable indicator that modern technologies can and should be used in conjunction with one another. You, too, can become a part of this story — open an issue in our GitHub repository, and we will be happy to make ODIN better together!
I hope that I was able to ignite your interest in trying Obsidian or even making a more tangible contribution to this incredible open-source community and thank you for reading this story about my journey.
I received two responses of people who actually produced something.
The vast majority, it seems, are professional note-taking junkies.
People in the Obsidian forum spend their lives fooling around with metadata conventions and "tagging" and "wikilinks" instead of producing work that actually matters.
Intellectual procrastination at its finest.
Now we have this... ODIN—a way to outsource your intellectual note-taking procrastination to AI.
Note: The requirement wasn't just a book. It was any "published knowledge product" which I define as book, code, course, essay, etc. (What Naval Ravikant calls Specific Knowledge).