The End of the Container Era: Apps Are a Historical Glitch

The way we use computers is fundamentally broken.

We accept it because we don’t have a name for it. We think the friction we feel—the fatigue, the scattered attention, the 32 tabs open at 4 PM—is just "work."

It isn't. It is a symptom of a specific architectural failure.

We are stuck in the Container Era.

The Historical Glitch

We assume that "Apps" and "Websites" are the natural state of software. We treat them like laws of physics. But they aren't.

Software containers are a "historical glitch". They didn't emerge because they are how humans think. They emerged because of macro-level constraints on how we used to build and ship software:

1. Distribution: Whether it was a floppy disk, an .exe file, or a SaaS domain, software had to be packaged into a standalone unit to be shipped, installed, or visited. We had to build "containers" just to get the logic to the user.

2. Servers: Vendors needed to ship one static, top-down interface to millions of users.

Because of these logistical hurdles, software became something "designed first and used second"—a fixed shape handed down to you. We are forced to hop between these rigid, server-defined boxes to get anything done.

The result is a system built on three anti-human pillars.

1. Siloed Logic

In the physical world, tools come to you. In the digital world, you must travel to the tool.

You have to "go to" Gmail to email. You have to "open" Salesforce to update a deal. You have to "launch" Zoom to talk.

These boxes don’t talk to each other. They are walled gardens of logic. You are not using a computer; you are visiting a series of disconnected rooms, over and over again.

2. Anti-Human Abstraction

This physical separation creates a deeper cognitive fracture.

I’ve been obsessed with this disconnect for years—I actually focused my grad project, Re, on it back in 2016. What I found then, and what remains true today, is that we are trapped in the "Page Era."

The "Page" is a centralized model where content and UI are pre-defined top-down by the server. The server decides what the container looks like.

But as humans, we don't think in Pages or Apps. We explore chunk-by-chunk, navigating threads of relevant info.

Think about it: You don't think "I need to open Google Calendar." You think "What's happening Tuesday?" But the OS doesn't understand that. It only understands "Launch Container A." It forces you to translate your intent into its rigid language.

The server ships Pages (grey boxes), but the mind follows Threads (red arrows). Visualizing the disconnect in Re (2016). View full slides on GitHub

Tabs are just the ultimate symptom of this mismatch. They are proof that the OS doesn't understand the thread of your work, so it forces you to hold the state in your head.

3. The Context Tax

Because the logic is siloed and the abstraction is wrong, we pay a price.

The Context Tax.

Since the boxes don't talk to each other, you have to be the connection. You are the human router, manually copying context from one box to another, 50 times a day.

The result is a quiet kind of insanity: 32 tabs open, 7 SaaS tools in the loop for one simple task, and infinite "where was that?" micro-searches.

You pay this tax with your attention, your time, and your nervous system.

The Shift Is Here

It feels like the dam is finally breaking. You can feel the energy shifting in the builder community.

Naval called it early with "UI is pre-AI". And I love what the team at AGI is doing—they are proving that the future isn't about navigating menus, it's about pure intent.

Then you have teams like Zo shipping the idea that "ideas are the new software" and unbundling the dev environment. You have Wabi exploring what personal software actually looks like when it fits the user, not the masses.

We are all circling the same truth: The "use" of software is the bottleneck, not the build.

It’s exciting to see this wave rising. But I think there’s one layer deeper we need to go. We don't just need agents that drive the old apps for us; we need to rethink the architecture so the "app" doesn't need to exist at all.

This is Part 1 of a series. Next up, I want to talk about why "Agents" are just a transition step, and introduce the "Faceless Data Layer."

(P.S. We’re cooking up something in stealth that tackles exactly this. Can’t wait to show you more soon.)

The Shift Is Here

It feels like the dam is finally breaking. You can feel the energy shifting in the builder community.

Naval Ravikant (@naval) called it early with "UI is pre-AI". And I love what the team at AGI Inc is doing; they are proving that the future isn't about navigating menus, it's about pure intent. Check out Div Garg (@divgarg)'s launch of AGI Mobile.

Then you have teams like Zo shipping the idea that "ideas are the new software" and unbundling the dev environment into a browser-native, personal intelligent cloud computer. You have Wabi AI exploring what personal software actually looks like when it fits the user, not the masses.

We are all circling the same truth: The "use" of software is the bottleneck, not the build.

https://x.com/amirhouieh/status/2007872633501860282?s=20

It’s exciting to see this wave rising. But I think there’s one layer deeper we need to go. We don't just need agents that drive the old apps for us; we need to rethink the architecture so the "app" doesn't need to exist at all.

This is Part 1 of a series. Next up, I want to talk about why "Agents" are just a transition step, and introduce the "Faceless Data Layer."

(P.S. We’re cooking up something in stealth that tackles exactly this. Can’t wait to show you more soon.)