Software for one is software a person builds with AI for their own use, and runs alongside their AI or their agent to get more done.
It will change what people count as the software market. Big software's moats won't last as long as people think.
The AI is there. So are the tools and the infrastructure to deploy and run what you build. This year you can build an app for your Mac or PC without much effort. You can build one for your phone and put it on your phone and a few friends' phones. More people will build the tools they need for whatever is going on in their lives.
Replacing the big tools
I've built software that replaces corporate tools for clients. CRMs, marketing systems, workflow tools. Almost every category. The standard objection is that you won't want to run it yourself. That misses the fact that AI also handles much of the work of keeping software running.
One client ran a specialty vertical CRM. We replaced it and tied the new one straight into their workflow tool, their quoting system and their production systems. The whole system fits their business, and they can adjust it whenever they need to.
We gave up 50 to 60 percent of the CRM's functionality. Most people use about half of what any software offers. So far nothing we left behind has been missed. It was functionality they never used.
What it takes
Building your own takes a few things most companies don't have yet. An appetite for change, and an organization flexible enough to act on it. People in house who can explain clearly what they want. And some patience. A couple of months, not the couple of years software projects used to take.
How it spreads
In the enterprise, it arrives as the skills spread. A manager gets tired of a line on their P&L and replaces it. Some enterprises will have rules against that, but it will creep in anyway, and I expect it to move upstream into most of the non-regulated parts of the business.
For individuals, it means less time fighting their tools and more room to shape how they work.
Where to start
The most important thing a leader can do is upskill their people. This is the human-first side of AI. Take your subject matter experts, the people who know your business and its processes, and teach them to build these tools. Bring in pure technologists from outside and you end up back in scoping discussions. What you want is smart, ambitious people who want to make a difference and are willing to try something.