I have been quite fond of the term “life logging” recently. The idea is simple: give an AI agent as much context about your life as possible. I think this is, for better or worse, how the future is going to unfold.
Let me explain why through my own experience.
Late-night context expansion.
1. The Context Expansion: A Personal Journey
My first encounter with AI-assisted coding was laughably narrow. I would copy an error message from my terminal, paste the suspicious lines of code into a chatbox, and ask: “what’s going on?” The AI saw a tiny sliver of one project — a constrained snippet with almost no surrounding context.
Then, at the start of this year, I began using Claude Code. It didn’t take long before I found myself putting the bigger picture of my project — the goals, the architecture, the plans — all into Claude’s memory. Why? Because it was the way to make Claude perform better, and it saved me from meticulously re-explaining the purpose of every session. Context was no longer disposable; it was reused across sessions. The global context became, essentially, everything I deemed useful for execution within one research project.
Now, I am increasingly aware that this approach has a natural ceiling. And the ceiling pushes outward in two directions:
Horizontal: across projects. I noticed that useful context isn’t confined to a single codebase. Best practices for codebase construction, cluster configuration rules, shared tooling conventions — these are relevant everywhere. The logical next step was to make cross-project context available to the agent. The scope expanded from one project to all projects in my research.
Vertical: beyond research. Once you start logging context for work, it feels oddly natural to extend it further — gym plans, reminders for tedious errands, personal schedules. The scope expanded from research activities to life activities.
And just like that, without any grand plan, I found myself doing “life logging.”
2. Two Cents on Where This Is Heading
2.1 Enjoy the Pre-Virtual-AGI World While You Still Can
Our workflows and lifestyles are changing rapidly, and I believe they will change so drastically by the time virtual AGI is achieved — which I don’t think will take very long — that today’s routines will feel quaint in hindsight. The way we interact with computers, manage our time, and even think about productivity is being quietly but fundamentally rewritten. If you find joy in the current way of doing things, savor it. It won’t last.
2.2 The Real Endgame: From Text Logging to Visual Logging
Seeing more than others can.
Now imagine real intelligence systems that can consume hours-long videos, understanding not just the words but the scenes, the objects, the spatial relationships, the intentions. Then imagine this linked with AR glasses — something like Meta’s Aria glasses.
This is the real life logging endgame: everything you see, uploaded to systems that understand what you are seeing. Not a text summary of your day, but a continuous visual stream processed by models with genuine world understanding. This, I believe, is the next game-changer product in the real-world AGI revolution.
And for that to work, you need real-world intelligence — in the sense of having internal world models that understand the physical world. Such intelligence would likely be more vision-first than language-first, a departure from the text-centric paradigm that dominates today’s AI landscape.
3. A Note on Trust
Companionship without words.
I am not saying any of this comes without security concerns. Giving an AI system a continuous feed of your visual experience raises profound questions about privacy, data ownership, and potential misuse. These are real, important issues.
But I have faith in humans. We have navigated the introduction of every transformative technology — electricity, the internet, smartphones — with stumbles and course corrections, yet ultimately arrived at something that works for society. I believe we will get there with AI too, and get there in a safe and secure way.
The future of life logging isn’t a dystopian nightmare. It’s a tool, and like all tools, its character will be shaped by the people who build and wield it. I’m optimistic about those people.