I spent 8 years building Google Sheets. Now I think apps are on their way out

I spent nearly eight years at Google as principal engineer for the Docs suite, growing Sheets from a five-person experiment to hundreds of millions of users. Building it took years, dozens of exceptionally talented engineers, and the kind of resources that were only available to the world’s biggest companies. Watching a small team spin up something functionally comparable in less than a week was, to put it mildly, clarifying.

In the age of AI coding agents, the standalone app has a problem.

When building an app takes days instead of years, the apps themselves are worth less. We’re already living an early version of this. My company uses a hiring platform for recruiting, but we barely touch its frontend anymore. We built our own interface on top of its API — one that better fits how our team actually works. The platform has effectively become a glorified database, and I don’t think we’ll renew.

That’s the shift founders need to pay attention to. Value in software is moving away from the interface and toward the data underneath it. An app frontend is increasingly a liability: opinionated in ways users didn’t choose, maintained on someone else’s timeline, and built for an average use case rather than yours. The founders who don’t see this coming will spend years building an interface that their customers will eventually replace themselves.

What fills the gap is something I think of as the “meta-app”: apps that build other apps on the fly, perfectly tailored to your immediate need. You tell it what you want, and it figures out the rest. AI coding tools like Claude Code and Codex are an early version of this. They’re getting a lot of attention for helping developers build software faster, but their real power is bigger than that. It’s the ability to go directly from intent to outcome, for any task, without an app in the middle.

My team used our own AI coding platform, Warp, to build the Sheets clone, and teams across the organization are using it to spin up personalized tools that replace the apps they used to rely on. Our creative director, who doesn’t know how to write code, uses it as his starting point for almost anything. He recently migrated our old website off its CMS just by describing what he wanted. He didn’t need an app to do that — just something that understood what he was trying to do.

The counter-argument I hear most often is that the meta-app still has to connect to existing systems somehow — through APIs, integrations, and connectors. That’s true today, but those layers exist because data is currently siloed across legacy apps. They are a transitional solution. The companies that structure their data for agents now, rather than locking it inside app interfaces, will be better positioned as that transition accelerates. At Warp, we store company data in simple files and databases rather than in SaaS products, because we don’t want to spend years doing integration work just to make our own information available to the agents that need it.

For any company still heavily invested in the current app model, there’s another problem worth paying attention to. Many SaaS companies will respond to this shift by trying to make it harder to get your data out, because the data they hold is increasingly their only defensible asset. Some will try to lock customers into their own agentic solutions, keeping anything you build dependent on their systems. That might buy them time, but it won’t change where things are headed.

Apps won’t disappear overnight. The transition will be uneven, and some categories will hold longer than others. Stripe isn’t going anywhere. Neither is Spotify. Companies with genuinely complex business logic, or with unique data at their core, aren’t going anywhere either. A frontend sitting on top of a database, even a sophisticated one, is a different story. The window for building a defensible software business around that kind of interface is closing faster than most people inside those businesses want to believe.

I realize this can sound threatening if you’ve spent years, as I have, thinking carefully about how to build great apps. But I think it’s actually the most exciting thing happening in software right now. We are moving toward a world where anyone can build what they need, just by describing what they want. We will all be software builders, even if none of us are writing code, and software will finally serve us, rather than the other way around.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.