Help me organize my time. And my money. And my attention, motivation and health.
So I’ve tried building about 7 life organizer apps, from managing my finances to building a work-life balance and everything outside of it. None of them solved the problem, and eventually, I started wondering whether the common denominator was me.
Micromanaging my own life is exhausting. There’s always something else to account for. There’s work, going to the gym, eating (healthy), and keeping the apartment clean. All of which fits neatly on a schedule. But then add the commute, groceries, expenses, health, motivation, relationship management, and the occasional need to do absolutely nothing, and suddenly the timetable looks less convincing. I even tried doing the opposite. Less structure, more flexibility, no clear plan for tomorrow. And it was terrible. All it did was increase my doomscrolling and fuel my procrastination, with a feeling of dread and a lack of motivation starting to mount above me.
So the next very reasonable thing to do was to organize every piece of time, money, and motivation management information into a coherent semantic space and use that to build out something useful. I’m talking every self-help book, psychology paper, and piece of time-management advice. Time to spend my entire weekend over-engineering my inability to wash the dishes (instead of just washing the dishes and getting over it).
I broke the problem into five components: money, time, attention, motivation, and health. I feel all of these are interconnected. Money means I can afford a cleaner and a chef, i.e., more time to go to the gym, i.e., better health. No money means spending more time working, finding jobs, passive income, investing, etc., which means less time to do other things, i.e., less fit. This may be an overly simple way of describing life, but it was a starting point. More importantly, we have five components to start us off with.
But then comes the next decision. The type of data source. There are books, research papers, YouTube videos, hyper-opinionated Facebook boomers, and many more sources of information. Quite frankly, it’s hard to say which is most useful. It may sound obvious to think, ‘Just listen to Warren Buffett for financial advice.’ Except I don’t have the capital to invest in long-term, low-return investments like Warren Buffett does. And I honestly read some pretty good advice from McDonald’s workers on Reddit. So all of these different realities had to be taken into account, giving our 5 components (money, time, attention, motivation, and health) another dimension — the source type: papers, books, articles, social media, and videos (transcripts).
5 components × 5 source types. Naturally, you think of vectors at this point, and a vector database is what I went with. I built a small prototype of this, collecting 20 data sources and representing each on my 25-dimension vector space. I also compared this to normal keyword search and open-source semantic embeddings. I connected this to a local Ollama model (Qwen) and honestly, the results weren’t bad. The chunks received from the semantic search were arguably much better, although the final result wasn’t complete. I needed to connect this with a harness and then apply that to various features.
The result was another project added to the pile. The reality was that the AI continued to do what I did: a perfectly solution-oriented approach with clean to-do lists and labeled spending. But the limiting factor was still me, the human. And that is the biggest takeaway from this. I can have access to all the world’s knowledge, and the fact of the matter is, humans don’t have the scope to deal with all of it at once. Additionally, perhaps, our innate nature for spontaneity and unpredictable experiences makes it less feasible for us to consistently follow through with AI’s production, which is programmed for a one-dimensional, constant, and perfect performance. This makes for an even more daunting question. Are we really holding AI back? Is our constant need for approving things and being the final decision-makers a type of setback? It may be necessary for now, as the systems are prone to errors. But what comes next, when they become so advanced that errors are no longer the case?
I started out asking how to organize my time. Now I’m wondering how much advice I can realistically absorb before the advice itself becomes another task. Somewhere between a laundry schedule and a vector representation, I started worrying about whether I’m ready for an AGI future. Am I overdramatizing? Probably. Am I avoiding doing the dishes by extending this article... absolutely.
Go wash the dishes.
Written by Ahmed Yar Khan. Edited, as always, by my sister.