I spent last weekend watching a woodworker friend do math in a notebook covered in sawdust, and I mean real math with fractions flying everywhere because he buys in board-feet and thinks in feet and inches and his client sends drawings in millimetres and his saw eats an eighth of an inch on every single cut.

I have never milled a plank in my life which is why his notebook hit me harder than any keynote this year, because I read his arithmetic as a layman, not as a craftsman, and the shock was how much of his day goes into conversions that a tiny program could do while he drinks his tea.

We have lived through this democratisation once already with hardware, which is why I keep thinking about dates. The mainframe in the seventies belonged to the company with the air-conditioned room, and then the Apple II in 1977 and the IBM PC in 1981 at something like fifteen hundred dollars put a computer on a desk, and then Jio in 2016 put data at something like three hundred and ninety nine rupees a month into a phone that costs less than seven thousand rupees in Tiruppur today. And the machine stopped being a privilege and started being furniture.

I grew up on Windows 98 and XP where the joy was in the decoration, and the themes and the widgets and the desktop toys let you arrange the machine the way your own head was arranged, and all of that slowly died because the industry decided there was one respectable way for software to look. So hardware got personal while software stayed wholesale, because building software still needed a team and six months and a budget, which meant every tool had to find ten thousand customers to justify itself.

That equation just broke in front of me last month on Omarchy, when I wanted a resolution switcher and told an agent to build me one and it existed in about ten minutes, and when I went to give it back I found somebody had already built the same thing. That ten minutes cost me something like forty cents in tokens, which is less than the coke I was holding. Its so sad to see how long we waited for this.

So here goes; my friend needs an imperial calculator for joinery which adds 7’ 3 5/16” and 5’ 11 9/16” and subtracts kerf three times and spits out cutting lists in both inches and millimetres with wastage baked in.

No sane startup would ever build that, because the total market is maybe twenty thousand woodworkers who still think in fractions and the support emails alone would kill you. And that is exactly the point, since he does not need a startup and he needs an afternoon with an agent that knows his saw blade and his supplier who sells in eight-foot lengths.

In plain English; the program remembers that his blade eats 1/8” so he does not have to.

[Tip: watch the notebook for a week before you automate anything, since the scribbles in the margin are the real spec.]

I keep collecting these one-person problems now that I know what to look for, and they are everywhere once you stop looking for unicorns. A hosiery trader in Tiruppur who thinks in dozens and bale rates and GST slabs and wants his billing in Tanglish the way he speaks it. A tailor master in Ludhiana who keeps blouse measurements in a diary with codes only she understands and loses a day every Diwali re-measuring regulars. A mistri in Kolkata who quotes in square feet but buys cement in bags and steel in kilos and does the translation on his phone calculator at midnight. Most times each of them adjusted themselves to Tally and Excel, which is generous software that never quite learned their language.

Wouldn’t it be nice if your tools fit you the way your tailor fits a shirt, instead of you squeezing into off-the-rack sizes forever?

Honestly, the part I have not figured out is what this flood will look like from inside, because right now there are three contenders sitting on my own machine. There is software which lingers as an app or a Quattro plugin that my friend can tap with dusty fingers, and there are prompts which evaporate after one chat and have to be re-explained every morning, and there are skills which travel as reusable know-how that turn a dumb agent into a joinery apprentice who already knows about kerf and board-feet.

I read the skills debate as a layman, not as an agent-infra person (or atleast that is how it seems to me), and my gut says the durable middle will win for boring reasons. Prompts rot because nobody remembers them, and full apps ossify because nobody maintains them, while a small skill with his blade width and his supplier lengths inside it can spawn a fresh calculator whenever he buys a new saw. Luckily the cost of trying all three has collapsed to near zero, so we will learn by littering, the way nature learns by generating and discarding.

The power question underneath is who gets to tax this, since the App Store took thirty percent for letting you in the door and SaaS took twenty nine dollars a month forever for a seat you barely sat in. Personal software dodges both by being too small to meter, which is also why the incumbents will try to meter the shovels instead with tokens and agent seats and skill stores. Technology always works for good and the leader always claims to be right, so watch where the rent moves next.

I want my work to have some impact and growth is how I increase it, and right now that means building tiny tools for tiny audiences without asking for permission. Work is worship, and the smallest workshop deserves its own software the way the smallest kitchen deserves its own knife.

Tweet to me the one-person software you would build for yourself this weekend. I am @troysk704.