I’ve been reading about this idea of “AI agent phones” where instead of opening apps, you just tell the AI what you want and it handles everything.

At first, it sounds like a natural next step.

But the more I think about it, the more questions come up from a development perspective:

• If AI is handling actions, what happens to traditional app UI?

• Are we moving towards backend-heavy systems with AI as the interface?

• How do we ensure reliability when AI is making decisions instead of users?

It feels like the problem is shifting from:
“how do users interact with apps”
to:
“how do we build systems that can act safely and consistently”

That’s a pretty big change.

I tried breaking this down in more detail here: The End of Apps? What the AI Agent Phone Is Really Pointing To

Curious how others are thinking about this. Are AI agents just another layer on top of apps, or are we slowly moving away from apps entirely?