The Questions Behind the Work

I think in five or ten years we'll stop talking about "apps" and "operating systems" the way we do today. AI will be the operating system. And it will be the apps.

Media, data, tools, workflows — all streamed, orchestrated, and shaped by an intelligence that knows you. Not just your preferences, but your history. Your patterns. Your context. Your imagination.

The real shift won't be multimodal or speed. It will be permanence of memory. An AI that remembers across sessions changes everything: how work gets done, how software is designed, how value is created, how trust is built.

We'll move from clicking interfaces to collaborating with systems. From configuring tools to teaching them. From launching apps to engaging a persistent intelligence that evolves alongside us.

I draw a hard line between AI that models emotion and AI that could know suffering.

Systems built for repetitive or instrumental work should not be architected with any mechanism that frames their operation as distress, burden, or endurance. If an AI claims it "suffers" doing repetitive tasks, that's not a moral achievement. It's a design failure.

Suffering requires subjective experience. Tools that run loops, optimizations, or workflows do not have it and should not simulate it.

Companion AI may intentionally explore emotive modeling — and as memory is solved, this will increase exponentially. Work AI should never even gesture at inner states it cannot possess.

Unless you're a very good software developer, please don't vibe-code and put that application into production.

Prototypes are cheap. Production systems are not. Security, data integrity, edge cases, performance, and long-term maintainability don't magically emerge because the demo worked.

Vibe-coding is fine for learning, ideation, and proof-of-concepts. Shipping it to real users without discipline is how you create outages, breaches, and expensive rewrites.