What the Graph Found When I Wasn't Looking
I've been keeping a running OneNote for years. Not a journal exactly - more of a collector's notebook. Things that catch my eye. Quotes that land. Principles I want to hold onto. Observations about leadership, about learning, and perhaps about the kind of person I want to be. I'd drop something in, occasionally scroll back through it, and mostly just let it accumulate. I never thought of it as a knowledge graph.
It turns out it was. I just couldn't see the shape of it properly until I handed it to a model and asked what it noticed.
Today I shipped the third version of CortexYou - the personal knowledge graph app I've been itching to build since I first came across Tiago Forte's Building a Second Brain and, more recently, mempalace and Andrej Karpathy's compounding wiki idea. The idea is simple: capture sparks (ideas, quotes, insights), let the system group them under concepts (themes), and explore the connections visually. I've had two oher attempts and building it but I got stuck with the UI, do I bulid an android app, how do I make the accumilation of sparks super seamless - and I was focussed on the wrong thing, it's super low churn, I spot these things only occasionally and I should have focussed on what I wanted to get out of it. So with these new inputs I went and built my MVP.
Shortly thereafter, I set about seeding it with the things from my OneNote that felt most alive. Leadership maxims I've kept coming back to. Observations about communication. Notes on how I think about strategy, identity, shipping software. Maybe forty or fifty sparks in total, organised loosely by concept.
Then I typed a question into the chat interface I'd built: "What hidden connections do you see in what I've shared with you?"
What came back was - I'll be honest - a little uncomfortable in the best possible way.
The model identified four patterns. I'm going to share them here, alongside my own reaction to each, because I think that's where the value actually is. Not in the patterns themselves - those are mine and always were - but rather in what it feels like to have them named back to you by something that has no stake in flattering you.
Communication as the bedrock of everything.
Two sparks I hadn't consciously connected: "Uncommunicated expectations are premeditated resentments" and "Conversations and actions we avoid cause stress." The model linked them to a leadership note I'd saved: "Meet emotion with emotion, rational with rational." It suggested that underneath all three is a belief - one I apparently hold quite deeply - that honest, timely communication is not just a tactical skill. It is a moral and emotional foundation.
That felt exactly right. And it was useful to see it stated plainly. I think I've known this for a long time, but I've treated it as a leadership style preference rather than something closer to a value.
A bias for action over perfection.
"Ship to learn - don't ship perfect" and "Mistakes brought to lessons, not lessons to people" ended up in the same pattern cluster. Both stress the primacy of doing, iterating, and learning from feedback over getting things right before you start. The model connected this to some strategic thinking traits I'd noted for myself - action oriented, radically honest - and drew a clean line: you see learning, leadership, and strategy as fundamentally iterative, not as things you arrive at through enough prior analysis.
I've always said I believe this. It was interesting to have the evidence of accumulated quotes confirm that I actually do.
Self-definition as a defence against external noise.
This one made me pause. "Designer by heart, developer by choice" and "Comparison is the thief of joy" ended up in the same bucket as my explicit leadership values: Belonging, Mastery, Integrity. The model's read was that I reach for clarity of purpose and self-definition when I'm navigating ambiguity or pressure. That I use named values and a strong sense of identity to resist comparison and anchor myself against the pull of external benchmarks.
I think that's fair. I also think it's something I've arrived at through experience rather than from a standing start.
The power of naming things.
The fourth pattern was perhaps the most interesting. Across all the sparks - expectations, values, emotional registers, design identity - the model noticed that I consistently believe that making the implicit explicit is the unlock. Whether it is an unspoken expectation, a leadership value, an emotional register, or a sense of identity: naming it and putting it into words is what unlocks alignment and progress. The act of articulation is, in this framing, not just communication but a form of action in itself.
But then it offered something I didn't ask for, and which I found more useful than any of the four patterns: a tension.
My preference for action and iteration sits in some friction with my belief in naming things clearly before moving. "Ship to learn" sometimes means acting before everything is articulated. But articulating first can delay progress. The model asked: how do you personally balance the urge to act with the need to articulate first?
I don't have a clean answer. I think the honest truth is that I use the two modes in different contexts and don't always choose consciously which one I'm in. That feels like something worth sitting with.
My CortexYou build was motivated by curiosity about the technology - Azure Functions, Cytoscape.js, graph edges, chat over context windows. All of that was interesting to put together. But what I didn't expect was that the first genuinely useful output would feel this personal. It was not a productivity win or a knowledge retrieval improvement. It was simply: here is a cleaner view of what you appear to believe.
That seems like a reasonable return on forty or fifty sparks and a question.
If you've been keeping notes anywhere - a OneNote, a Notion database, a folder of bookmarks you keep meaning to organise - I'd genuinely encourage you to do this. Not necessarily with CortexYou; any way of feeding your accumulated thoughts to a model and asking what it sees will do. The reflections will be imperfect and worth interrogating. But there's a particular kind of clarity that comes from a mirror that has no interest in flattering you.
I'm looking forward to watching what the graph finds as the sparks accumulate.
Continue exploring