Essay
Executive function is a systems problem, not a character problem
I was diagnosed autistic in early 2025. I had known about the ADHD for years. What the autism diagnosis changed was not my behavior. It was the explanation I had been using for it. A career of ideas that arrived whole, and Tuesdays that disappeared. A house with my husband and three boys where the routine kept landing on me, and a private theory that other people were simply better at remembering. They were not better. They were running a different process, and I had spent twenty years trying to install theirs.
The advice was always some version of the same system. Break the task down. Get a planner. Review it weekly. Keep the list trustworthy. All reasonable. All of it had the same hidden requirement: remember to use it. That is the exact faculty I do not have reliable access to. By the time I noticed the system had lapsed, the cost of restarting it was higher than the cost of the original mess, so I didn’t. I called that laziness for a long time. It was a context failure. Between one day and the next the working set evaporates, and rebuilding it costs more than the task did.
What does work, for me, is narrow. Spend the day on the things I am actually good at. Get the remembering and the routine off my plate. Do not ask me to initiate the boring part, because initiation is the expensive step and the boring part is where I stall. I did not have a way to do that until the diagnosis and the coding tools arrived together. One gave me the lens. The other gave me hands. I could take a specific failure from my own week and build software that does not get tired, bored, or overwhelmed, and that does not need me to remember it exists.
What I think transfers
The apps are for one household. I would not hand them to anyone. Three rules under them might be worth stealing, including if you never write a line of code. A spreadsheet with these properties would have beaten every planner I was given.
Never ask the same question twice. If a decision was made once, it belongs in a ledger, not in my head. YarMeals dropped from about forty questions a shop to four this way. What are we eating, given what we have decided before. The change had nothing to do with the model. It was the decision not being re-made.
The system starts the work, not me. The daily shortlist exists before I sit down. The documents file themselves. The shop begins with a cart already loaded. My job is to react, which I can do. A tool that waits for me to open it is a tool I will not open on the day I most need it.
Everything is reversible and logged. I could not trust software with dinner, mail, and paperwork until I could see what it did and undo it. That is an accessibility requirement dressed up as engineering hygiene. If a system can silently do the wrong thing, I will not let it do the right thing either.
What I would not copy
The first version of all of this fell over the moment it met a real week. Agents invent. They file the document under a confident wrong name. They will happily automate a decision I had not actually made. The useful constraint was making the undo cheap, and refusing to let the tool ask me a question it could have looked up.
It also did not fix the rest. I still lose the thread. I still do a bad job of networking, which is part of why this site exists instead of a healthier LinkedIn habit. The apps cover four drains: meals, documents, mail, the next thing. They do not cover a life. On a bad week the accommodation is that the cart is already loaded, not that I feel finished.
I’m not selling a method. I’m writing it down because writing is how I remember, and because most of what I was handed was coping advice with no example of the accommodation built into the thing itself. If one of these rules saves someone a year of trying harder, that is a good trade. The apps are there if you want to see the rules under load, including the parts that broke.