When AI becomes the user

Most SaaS businesses were built around people using screens. The interface organised the work, the workflow held the customer, and more users meant more paid seats.

That model changes when an AI agent can retrieve context, update systems and carry out the next step itself. The interface still matters, but increasingly as a place for oversight and exceptions. It becomes harder to defend the same price for a layer that customers use less directly.

I expect more software pricing to follow consumption, access and results. The strategic question is where the enduring value sits: the data, the permissions, the business logic and the infrastructure that lets actions happen reliably.

Adding chat to an old product leaves that question unanswered. Agents need understandable data, usable capabilities and a way to operate across systems without a person translating every step.

Existing customers and contracts buy time. The useful work is deciding what to do with it.

The opportunity belongs to businesses that make their underlying systems genuinely useful to AI.