Summary: What Re-Reading Taught Me, and What's Ahead
A fifth re-read of Atomic Habits: why long- and short-term goals conflict, and how habit stacking, retros, and identity echo software and AI-agent system design.
This is probably the fourth or fifth time I’ve reorganized my Atomic Habits notes. As life experience piles up, each pass gives me a slightly different read. What hits me most this time echoes a line from the book: “long-term and short-term goals are almost always in conflict.”
Most people, and most companies, live and survive in a way that needs fast feedback. There’s nothing wrong with that, but from where I stand it’s still a pity, because it quietly costs you some possibilities.
That doesn’t shake who “I” am, though: my own identity. I’m a long-term person: someone who, while still knowing how to enjoy the present (boredom, and facing challenge and fear included), keeps consciously iterating on my own life system.
Just like my 2026 yearly goal: build a leverage system I can use for the next five years, one that uses AI, knowledge, skill, and relationships to multiply the returns several times over.
And like what I found after moving into software and doing Agile: the thing I think matters most is the retrospective. The “good to great” chapters of Atomic Habits stress the same thing.
Where It Echoes System and AI-Agent Design
There’s a lot in here that echoes software system-design principles, even the way we design AI agents today. For example, habit stacking maps onto event-driven architecture in system design (hooks, brokers, and so on).
I catch myself carrying these habits into AI-agent design without noticing, and that’s when I realized: a lot of my harnesses run well precisely because the boundaries are set well. Event triggers, minimal-context feedback, real-time adjustment, all of it matters.
Like the book says, everyone has the goal, everyone wants an AI agent they can command. But what actually makes it work is the system details inside, including the ones the internet keeps repeating (lint, test, smoke test), so I’m especially demanding about those.
At the start I did hit walls everywhere and got told I was wasting time; I even doubted myself for a moment. Luckily, after five or six months of accumulation, I can now run automation confidently and effectively, and intelligently let it choose for itself, multiplying my productivity.
Design Your Own Life System First
I’ll share more on system design, AI-agent design, and how they echo each other when I get the chance. But what I want to say is this: we might as well design our own “life” system well first, because that is the key to everything. Once we have these habits, we naturally get clearer about what we want our agents to become. In a time that keeps changing fast, holding your identity steady is what keeps you from drowning in information, and it lets you ride the tailwind of this era.
And the next few pieces are where I actually land all this onto my own tools. Some are still running, none published yet, so consider them a to-be-continued I’ll fill in over time: a habit-tracking setup in TickTick, an “inevitability” environment built with a handover plugin, and splitting context across separate YouTube accounts.
References: Atomic Habits, James Clear
Related: back to the whole-book map; or revisit Systems Beat Goals, Identity: Who You Are Drives What You Do, Reflection and the Rigidity Trap, and The Goldilocks Rule and Falling in Love With Boredom.