For the last decade, software has been trying to turn every human activity into a workspace.
Your notes became a collaborative knowledge base.
Your calendar became a productivity system.
Your habits became a dashboard.
Your hobbies became content pipelines.
Everything became a SaaS onboarding flow.
Some of that software is good. A lot of it is useful. But the default shape of software got bigger than the problems it was supposed to solve.
Then AI changed the economics of small software.
Not because it makes everyone a founder.
Because it makes weird, personal, one-off tools worth building again.
That is the part I care about.
Not "everyone can launch a micro-SaaS."
Not "every workflow is a company."
Not "every script needs a landing page."
The interesting part is not that every personal tool can become a product.
The interesting part is that it no longer has to.
A tool for one commute.
A tracker for one car build.
A script for one annoying travel workflow.
A dashboard for one old tablet sitting on one desk.
A tiny utility that answers one question better than the full app does.
None of these are billion-dollar ideas.
That is the point.
For a while, that kind of software was hard to justify. Building anything required too much setup. You had to write the code, design the UI, host it, connect the APIs, debug everything, and maintain it. Even small ideas came with annoying overhead.
So you either used whatever app already existed, or you did the thing manually.
AI changes that math.
It lowers the cost of making software that is too specific to justify otherwise.
That matters because a lot of useful software is specific. Annoyingly specific. Embarrassingly specific. Specific to your apartment, your commute, your car, your job, your habits, your calendar, your family, your way of thinking.
That kind of software does not always make sense as a product.
But it makes a lot of sense as a tool.
There is a difference.
A product needs a market. A tool needs a job.
A product needs onboarding. A tool can assume context.
A product needs to explain itself. A tool can be built around the fact that you already know why it exists.
That is what makes personal software feel different. It can skip the generic parts.
It does not need to support every use case. It does not need a settings page for every edge case. It does not need team management, export flows, notification preferences, billing, or a growth loop.
It can be narrow.
It can be opinionated.
It can fit the actual moment.
This feels like a return to what personal computers were supposed to be good at: making the machine adapt to the person.
Somewhere along the way, a lot of software started asking the person to adapt to the machine. Learn the system. Set up the workspace. Configure the dashboard. Maintain the workflow. Keep the tool updated. Feed the app enough data so it can eventually become useful.
That is fine for some things.
But not for everything.
Some software should just do the thing.
The next wave of personal software probably will not look like one perfect app. It will look messier than that. A folder of scripts. A few Shortcuts. A local dashboard. A small MCP server. A personal site. A couple of agents with narrow jobs. A weird little interface for a device that was never meant to run it.
That messiness is part of the appeal.
Personal software does not need to be universal to be valuable. It just needs to fit.
And now, for the first time in a while, building software for one person is starting to make sense again.