A colander and a bowl both hold food, so it seems reasonable to want one object that does both. But a colander drains and a bowl holds, and those two jobs undo each other. Combine them and you get something that cannot drain properly or hold liquid without leaking. Add a toggle to switch between the two and it is only more complex, more expensive, and worse the day the toggle breaks.
It is not about a smarter object
The answer is not to make the object smarter, and it is not always to split it in two. A well-made one-piece bowl can pour off water when you tilt it, without spilling its contents, and still takes up the space of a single bowl. Or you keep a detachable colander or strainer that drops into any ordinary bowl you already own. What that takes is being clear about the job it is for, not cramming in another feature.
The clutter just upgraded
Software gets asked to be that combined object all the time, and the pull is stronger now that a tool can do almost anything. All-in-one, one-stop shop, super app, and now AI that (almost) does it all. It is the same wish as the kitchen, one thing to hold everything so you never have to reach for anything else. We used to keep a hundred browser tabs open, ideas and inspiration we meant to come back to (guilty). Now the same late-night itch reopens as a spark of curiosity in a generative tool like Claude, one that will think with you and even build you something shaped to you. The clutter got an upgrade.
Somewhere in the pile, no, somewhere in the chat, the way it happens in any ordinary conversation, it gets easy to forget the job you were actually solving for: the clever one-piece strainer bowl now collecting dust, the tab you never reopened, the chat you cannot find again, the context doc you forgot you wrote, the search term that never quite lands, the word on the tip of your tongue that will not come. The feature that made sense on its own, and quietly made everything else harder.
So what now?
There is no shortage of method for getting it right, and no shortage of argument over what "right" even means. Jobs to be done, first principles, the minimum viable version, various design thinking models, less but better. I am not loyal to any of them, and you do not need to be either. Underneath the vocabulary, each one is just a way of asking the same small questions: what is this actually for? Who is it for? Why? So what?
Once a tool can do almost anything, the hard part is no longer what you can add, it is what you are clear enough to leave out. A thing that tries to do everything rarely does any one part well enough to disappear into your work, and the tools people keep are the ones that know their job. A good bowl is content to be a bowl, and a good colander drains and stops there. Design the right thing, not every thing, and let the rest go.
The trouble is that "right" is rarely that clean once you are working with a team. In B2B or enterprise there are more people, more truths, and more history to work around, each holding a piece of the picture and none of them the whole, and each wanting to add a spout, a lid, one more thing, until you are back to the bowl that cannot hold water. That is design by committee, its own hard problem, and an essay for another day.