3 weeks of perfect runs - my personal system is finally humming

🔥

This is the system I'm proudest of, and it's deliberately worth nothing to anyone but me. I've been building it for about two and a half months. My goal? Free up time personally, and test and learn everything I was learning here so that I could be confident in building out things that are proper products at work (or, more accurately, architecting things for my team to build). So this system flies in the face of some of what we learn - it's not meant to have a product launch, it's meant to be so damn unique and customised that it holds no value to anyone other than me.

Current outcomes (and I'm going to build again so this will improve):

~63 hours per year freed up for me
~ 104 hours per year freed up for my husband
~$2400 per year saved in food costs (around $50 per week approximately) and reduced food wastage
A calmer family who are not stressed out about what needs to be done when for school - we've gone from a last minute scramble or forgotten gear about once every 2-3 weeks to one item that was a day late in the last two and a half months. (With the root cause of that miss fully addressed instantly in the system).
Three weekly runs in a row where I didn't have to change anything in the heaviest and most challenging workflow (the weekly planning workflow). I just reviewed and approved it to write.

Every minute of build time has already been paid back with interest before counting the time freed up for my husband, and I'm only two and a half months in. In fact, the first thing I built paid back on the first run with me being an hour better off from that first run even when you factored in the build time.

Here's how it's actually doing that:

Term-start — the first thing I built, and the one I love most. It takes a 7 day timetable screenshot and a CSV file exported from the school app and turns it into an ICS file I can upload into my calendar. Rather than manually creating it, it can apply the reasoning required over the root information and give me the output. It even has a second-pass logic built in for the edge case logic where days aren't in the CSV which usually happens for 1-3 days each term. It works flawlessly. It began as a series of saved prompts and was my first ICM pipeline once I found this community.

School comms - because that's honestly a real struggle some weeks to stay on top of while also being present for my family and working a fairly serious job. I have async capture based on classic email filters and a google app script that runs each day to write everything with those filters across into a file ready to be picked up. I also have a dropbox for screenshots from the school app or text copied and pasted from the parents WhatsApp group. All of that is synthesised into themed notes (fresh ones for each term, old ones get archived). Which means it's all natural language queryable and I get a weekly brief each week about what's going on that week that we need to know (more importantly, that's a handoff and input to the next stage - the weekly planning - which means I don't actually need to pay much attention to it at all most of the time). It picked up an error in days from what the teacher said in a class announcement straight away - and yes, the teacher was actually wrong and had to send out a class-wide correction when I contacted her to clarify.

Weekly planning - this is the tough one, the one that used to take up a bunch of headspace and effort. It's all about constraints, rhythms and logic - there's the usual patterns each week and then all the things about each week that don't follow them and knowing what to relax. I understood these trade-offs intimately - I was applying them each week - but fine-tuning those into a structure that meant Claude could run it took time and iterations. Config files that I didn't realise were needed until the right exception came up. Figuring out that daylight savings logic needed to be captured in a config table because the UTC<>NZ timezone conversion was biting me and breaking what had been working when the clocks changed. Tightening and relaxing constraints until the reasoning was consistent week to week with what I would do myself. When I started, this could only run on Opus and even then with a lot of hand-holding and guidance about those judgement calls, Sonnet couldn't even scratch the surface and was totally outmatched by the reasoning required. Now that I've got it refined, fine-tuned and ICM-ed to the hilt? It runs best on Sonnet (adaptive) - without the (adaptive), Sonnet is too conservative and needs way more input from me on the decisions, which means I'm not delegating as much of the work as I want to. Now that the system is nailing it I can do other admin tasks while it runs, which is the main way I get my time payback each year.
The tail end of the weekly planning workflow is a handoff that supports meal planning and shopping list generation. Weekly planning instructs on how many meals and what kind - 1 night versus 2 night cooks, anything that has to be quick enough to cook after school - and meal planning balances for variety, time and effort based on the brief. The shopping list then gets information from me about what's on hand and creates two shopping lists - one for the supermarket and, because I work for an FMCG company with a staff purchasing programme, one for me to use to make my work purchasing easy. (And by the way, that meal planning and shopping list generation, including staff purchasing, has been published organisation-wide for all staff to customise with their personal recipes and use. So I guess there was a tiny generic product hiding in there under all the personalisation after all).

Some choices that might be of interest:

I call my system one that is specialist based rather than agent based - it feels more honest and true to what it is. And most of the time, I don't need a specialist, so I don't bother (build only what adds value). Where it does pay back is where it codifies a judgement or behavioural stance, so my files there are pretty light and contain that only. Anything else goes where it belongs according to ICM - rules about what to do or not do means governance; instructions on the task means it goes in a skill file; ongoing context belongs in a config file. I'm pretty brutal, so these files are light and that's how they work best. Two specialists for my personal system - my family assistant and my school expert. That's all that it needs.
I've got a lovely string of mums who'd dearly love the school comms system for themselves. And I'm consciously not building anything scalable or saleable here (though I have taught a couple of them how to build one for themselves). The methodology, absolutely — that's going where it earns its place, at work, for products there. And the bits that apply to everyone's life with a bit of customisation - meal planning and shopping lists - that part could have the serial numbers shaved off and some materials to customise it and be shipped as a small working piece. But this whole system, as it is? It's mine. Its whole value is that it's so completely shaped around my family that it's worth an enormous amount to me and absolutely nothing to anyone else. You couldn't drop it onto another family if you tried. It solves what's true for my life; it's customised, individual, built entirely around my values and the way I like to do things.

That's the whole point. Above everything else, the value of this system is that it's mine.

Part 1 of 2. The work one's coming next. More detail on what the personal system actually looks like is in the attached image, and I'd love any questions, comments or feedback on this.

42 comments

3 weeks of perfect runs - my personal system is finally humming