For twenty years, software companies trained us to believe the product was the interface.
We bought seats. We learned dashboards. We created workflows. We hired people who knew where the buttons were.
AI changes that assumption.
The interface is becoming optional
If an agent can understand an objective, use the software, and return the result, the human no longer needs to become proficient in every application underneath the work.
That does not make software disappear. It makes software more infrastructural.
The next software company may have two customers
One is the human who owns the outcome. The other is the AI that actually operates the product.
Those customers want different things. Humans want confidence, control, and evidence. Agents want structured actions, reliable permissions, clean data, and predictable interfaces.
Features become capabilities
In SaaS, a feature is something a user discovers and operates. In post-SaaS, the same feature becomes a capability an agent can invoke when needed.
The product question changes from “how do we get users to click this?” to “how do we make this capability safe and useful inside a larger objective?”
I do not think SaaS ends. I think the center of gravity moves. Software becomes less like a destination and more like an operating layer.
