Software was never made for you. That's about to change.

There is a spreadsheet somewhere in your business, held together by four formulas and a colour code that one person understands, quietly doing a job that a real software company would charge you nine hundred a month to do worse.

Nobody planned it. It grew, the way ivy grows. And the reason it exists is the most interesting thing about it: the software you actually needed didn't exist, so somebody built it out of the only material they had within reach.

Which means the truth is already sitting in your business, disguised as a bad habit. Almost every company on earth is writing custom software. They just call it a spreadsheet, and they apologize for it.

Almost every company is already writing custom software. They just call it a spreadsheet, and apologize for it.

To understand why we ended up here, it helps to notice the single economic fact that has shaped every piece of software you have ever used. Software was expensive to make and almost free to copy. That's it. That's the whole story. If building something costs a fortune and duplicating it costs nothing, there is only one rational move: build one thing, and sell it to as many people as you possibly can.

And to sell to everyone, you have to build for the average customer. Who, famously, does not exist.

The US Air Force learned this the hard way in 1950. Their pilots kept crashing planes that were, by every measurement, working perfectly. So they measured four thousand-odd pilots across ten physical dimensions to refine the cockpit around the average man. The number of pilots who were average on all ten came back as zero. Not a small number. Zero. The cockpit had been designed with great precision for nobody at all, and the fix wasn't a better average — it was the adjustable seat.

Most business software is still that cockpit. It fits the composite of ten thousand companies and therefore fits none of them, which is why every tool you own has a settings page. A settings page is not a feature. It's an apology — a small, polite admission that the product knew it was guessing.

You've felt this even if you've never named it. It's the onboarding question that asks "which of these best describes your team?" when the honest answer has never once been on the list. It's the CRM built for a twelve-person sales floor being operated by someone who doesn't have a pipeline, they have about forty people they genuinely like. It's the eleven browser tabs, three of which are subscriptions nobody can cancel because they're attached to the personal Gmail of a person who left in 2023.

So we adapt. We bend the business to fit the tool, and we call it best practice, because we bought a thing and now we work the way it wants is a harder sentence to put in a board pack.

You don't just buy software. You buy its opinions about how a business like yours should behave.

That trade was never irrational. It was arithmetic. Building something shaped exactly to your business meant a development team, a discovery phase, a timeline measured in quarters and a budget that only made sense above a certain size. So the sensible thing was to rent someone else's assumptions and get on with the actual work.

The arithmetic is what changed. Not the ambition, not the appetite — the arithmetic.

Because the expensive part of software was never the idea. It was the translation. Months of patient, costly work turning what you meant into something a machine would agree to do, conducted in a language you don't speak, through people who charge by the hour to speak it for you. That translation layer is the thing AI has quietly demolished. What used to be a quarter is now an afternoon, and what used to require a team now requires a clear head and someone who knows what they're doing.

And when the cost of making something falls far enough, a category that never made sense suddenly does: software with an audience of one.

Software for one person isn't a smaller version of software for a million. It's a different thing entirely — the way a tailored jacket is not a small tent.

This is the part worth thinking about properly, because it's easy to hear "personal software" and picture a slightly nicer to-do list. It's bigger than that. It's a personal CRM that doesn't care about lead scoring, and instead tells you that you haven't spoken to someone in seven months and that you promised to introduce them to somebody. It's a studio's project tracker that knows the five stages that studio actually uses, rather than the fourteen an agile consultant once insisted on. It's a restaurant that knows what Tuesdays do. It's a clinic whose booking system understands its own quirks instead of quietly demanding the clinic get less quirky.

None of those are products. That's the point. Nobody was ever going to build them, because there was never a market — there was only ever a business, and a way it works, and the gap between the two.

What arrives with that shift is something we haven't had in computing for a very long time: ownership. Not the licence-agreement kind. The real kind. Your process stays yours and the tool bends around it. "The system won't let us" stops being a sentence anyone in the building has to say. The knowledge of how your business actually runs stops leaking into a vendor's roadmap and starts compounding inside your own walls.

For most of computing history, the individual has been the user — a lovely industry word for someone with no vote. Personal software makes you the author instead. It hands back a kind of authority over your own working life that most people gave up so long ago they've forgotten it was ever on the table.

For most of computing history we've been users — a polite word for someone with no vote.

This is where I'm meant to tell you it's all frictionless from here, so let me not. What has dropped is the floor, not the ceiling. It has never been easier to build the wrong thing beautifully, at speed, in an afternoon — and the next few years are going to fill up with elegantly engineered answers to questions nobody asked. Software also still has to be fed: hosted, secured, understood by someone after the person who made it moves on. A tool nobody owns becomes a haunted house remarkably quickly.

So the scarce thing has moved. It used to be the ability to build. Now it's the clarity to know what deserves to exist — which parts of how you work are genuinely yours and worth protecting, and which parts are just habits you inherited from a piece of software you no longer respect.

That's not a technology question. It's the same question that sits under a brand: what is this business, actually, and what does it need to become? Answer that honestly and the software almost specifies itself. Skip it, and you'll simply generate your confusion faster than before.

The businesses that pull ahead in this decade won't be the ones producing the most software. They'll be the ones clear enough about how they really work to shape something around it — and confident enough to stop apologizing for being a shape the market never made.

You already know your version of this. It's the spreadsheet with the colour code. It's the workaround everyone learned instead of fixing. It's the sentence that starts "if only it could just—" that you've said out loud, more than once, this year and last. That sentence isn't a complaint. It's a specification.

If you've got one of those — the idea you've never raised because it felt too specific, too small, too you to ever be built — that's precisely the kind of thing worth a conversation. Tell us about it. The specific ones are our favourites, and they're the ones that change how a business feels to run.

The age of software made for everyone is ending. What comes next is software made for you — and the only real question left is what you'd build if the answer could finally be exactly this.