Tag: personal OS

  • My Personal OS: Many Projects, One View on All of It

    For years my personal productivity system was Trello. I wrote about it in 2020: a board for life admin, a board for the recurring stuff, and a card each week asking whether I was actually making time for my priorities. Work was in Asana, because at work you have to interop with everyone else.

    One system for your life and another for work makes sense when you have a job. A portfolio life is different. Life is work, work is life and all the lines get blurred. Different workstreams live in different places, shared with different people. Add AI and there are a lot more moving pieces and things to keep in sync. But – I’m one person, and I need one view on it all.

    So earlier this year I moved everything into a git repo: plain files that Claude Code keeps up to date, and that I can still read. The work itself still lives wherever it’s shared – the repo reads from my inbox, calendar, GitHub and the rest, so I get one view without asking anyone else to change tools. I started with this and it’s been getting progressively more elaborate ever since.

    What it does

    It keeps me oriented and on track against the things I’ve decided are important. Every month I set some goals and every week I make a plan.

    Diagram: the working loop, from monthly goals to a weekly plan to each day

    Then each morning I say “run the checks” and it goes through my inbox, calendar, GitHub, course sales and what I’m waiting on, then tells me what matters. During the day I throw things at it (“the 2pm cancelled”, “release is done”) and it updates the week’s plan. It has the standard guards: it drafts emails but never sends them without a yes, and it never deletes anything.

    Nothing runs on a schedule. If I don’t open a session, nothing happens. I might add a morning cron so that when I start my day the checks are in, but also it’s fine. The ritual of writing “hi it’s [day]” is a moment not just for Claude, but also for me.

    Diagram: what the morning checks look at, including inbox, calendar, GitHub, course sales and what I'm waiting on

    At the end of the day it pulls in my notes and we write the wrap-up. Once a week we check in on strategy. Portfolio: where my time actually went against what I said my priorities were. Content: how my social media performed, and what it drove.

    I kept asking “give me the burndown” so now that’s on a page snapshotting my day: the burndown (I tick things off there and it picks them up), calendar, the week, what needs me, and any decisions I haven’t made yet. Decisions were getting lost in chat, so this was a useful improvement.

    A hedgerow

    Then Ellen Chisa wrote up her personal OS. Ellen’s feels designed. Mine feels… organic. It works, and I love it. It also grew one annoyance at a time, so it had the architecture of a hedgerow.

    Which is fine, for a while. Something goes wrong, you add a rule wherever you happen to be, and tomorrow goes better. It’s only when you try to change something that you notice a rule like trying to keep to a 30h week is written down in five different places, and you just updated one of them.

    So I asked mine to read her post and tell me what it was missing. No metric. No log of what it actually ran. No glossary (it asks me who “B” is roughly weekly). It also pointed out that these would have caught problems that had gone unnoticed for too long, like daily checks that had quietly stopped running for weeks. I said I’d take all of it.

    Then, I added some structure. A router file that just says where things live. Each rule written down in one place, with the date it was added. A line at the end of every run saying what ran (ran: inbox ✓ · money ✗).

    The end-of-day wrap-up already asked how the day went, but it was too vibes; now it tracks must-dos done against planned, what came in unplanned, and how I’m feeling. And from Natasha, a hook that opens every session by telling me whether today’s checks have run. None of it changes what the system does. It does mean that a diagram can now capture and explain it.

    Diagram: how it's organised, with one router file pointing to rules, checks, playbooks and reference, and the state in plain files in git

    In the end, I love this point from Hilary Gridley: “Ultimately you don’t just want the tasks organized, you want them done.” For me to get things done I need a level of orientation, and not to be worried I’m missing something. Once I have that, I can prioritize and burn down.