System Design - User POV
Building a Personal Knowledge Hub
Sep 2026
The idea began with a simple question: "What if we could spot patterns, revisit old ideas and connect new ones?"
I was looking for ways to extract more value from the information I had saved over the years. It was not just about retrieval but about context, and the connections that lie below the surface. It was not just a question of “How do I save more?”, but “How do I make what I have already saved more useful?”
Now, I have always been the kind of person who saves things. Not in the dramatic, obsessive sense, but in the way that people accumulate useful fragments of their lives - links, notes, references, half-finished ideas and things that seemed interesting enough to save and retrieve later. Over time, that material had spread across different platforms... Text notes in the cloud, a growing collection of bookmarks in Raindrop, good YouTube videos saved in multiple YT playlists, and more.
For a while, I was patching the problem in all the obvious ways by tagging a few things, saving summaries as needed, and adding notes next to references that seemed useful. Those were all valuable habits, but they were still just patches after all.
The real issue was not scarcity, it was fragmentation. Yes, a saved item was usually saved in cloud-based services and available across devices, but the context around it could be hard to recover. That is how the idea for 'ThinkHub' began to form.
Saving is easy. Finding is hard.
"What if I didn't aim for a cleaner archive, but a better way to reconnect to an idea?"
The more I thought about it in those terms, the more the limits began to fade away. Personally, it would also be a way for me to gain intensive, hands-on experience with AI-assisted software development via a live project (instead of creating random hypotheticals or just reading about it in theory).
So, I started using AI tools to address that challenge head-on. Initially, it was just a few useful scripts that allowed some level of automation and cleanup. However, as the momentum built-up and I gained a foothold in this brave, new world of programming, the project soon started taking a life of its own.
The first real benefit was simple and satisfying: I could now find things with far less friction than before. If I saved something useful, it no longer felt buried inside a specific service, but was available across multiple datasets. I could revisit it as needed, and engage with the material faster than ever before.
That mattered more than it sounds like it should. Soon, the archive started feeling more generous. I no longer had to rely on my memory to keep the useful parts alive, and so much more of the older material became usable again!
Then came the stage of adding more intelligence to the system by infusing LLM processing, auto generating insights and actionables, and adding importance scores and relevant tags to each item. With those engines firing, the value of the data increased even more.
A saved article was no longer just a source. It became a memory hook, a reference point, a node in a larger web of patterns. A piece of text gained more value when it sat next to related material. A topic became easier to understand when the items around it were connected instead of isolated. That was the moment the project started transforming from a storage and retrieval tool to a knowledge driven hub.
It began to feel like a "system"
The timeline is not dramatic, yet meaningful.
In June, I began by focusing the project on 15+ years of saved notes, bookmarks and references. The first few workflows may not have been too ambitious, but were enough to make those archives easier to revisit.
By early July, it was not about saving more, but about revealing the underlying context that makes them useful. Auto enrichment of summaries, tags and metadata began to matter, because that helped keep the meaning alive.
By mid-July, the whole thing started to feel more like an integrated machine, as related items began surfacing connections I had forgotten about. That was also when the real shift happened: I now had a working toolkit using which I could save multiple types of information from disparate sources, enrich them suitably, and retrieve them in a meaningful way.
By August, the 'ThinkHub' project started taking the shape and form of a "knowledge management system". It was now easier to move through all the material, revisit paths and discover new relationships. Honestly, it felt more like thinking with my archive instead of struggling with it.
The benefits package
Google Sheets for backup and lightweight indexing, Raindrop for saved links and references, Obsidian for personal note-taking and connected thinking, YouTube and other sources for material I wanted to revisit later, and LLM processing to turn raw content into summaries and context. The mindmap scaffolding I built on top of it, added yet another layer for visual and text-based ways to explore relationships across the entire dataset.
The quality of the integrated experience had now changed significantly. Search became faster, discovery became easier and connections between topics became more visible. Instead of treating my archive as a collection of things I had once saved, I began to treat it as a living layer of context I could navigate.
And, that is where the real value of 'ThinkHub' delivered. What began as a collection of Python scripts in a disjoint workflow was now a neat package of benefits that spanned multiple data sources and information types.
The system as a whole becomes far more useful when the limits of any one part disappear.
What lies ahead...
The project still has a few rough edges, but they are no longer the most interesting part of the story.
The roadmap for the next few months is clear: More sources are planned to be integrated, more ways to visualize the material are to come, and more of the dataset will soon be connected.
The 'ThinkHub' project is now on its way to becoming a personal knowledge system that keeps growing in breadth, clarity and usefulness.
System Design - Engineering POV - A developer's view of this project
This project captures ideas, decisions, and lessons I’ve learned while building with AI. It aims to help non-technical users understand and use these tools effectively. I also use AI at times to improve or refine the content.
Readers should use the commands and instructions here with care. The creator accepts no responsibility for system damage, data loss or other consequences resulting from their use.