Lately, my group constructed a working clone of Google Sheets in just a few days. It’s inferior to the true factor, however on the present tempo of AI, it may very well be quickly sufficient to matter.
I spent practically eight years at Google as principal engineer for the Docs suite, rising Sheets from a five-person experiment to a whole lot of thousands and thousands of customers. Constructing it took years, dozens of exceptionally proficient engineers, and the type of assets that have been solely out there to the world’s largest firms. Watching a small group spin up one thing functionally comparable in lower than every week was, to place it mildly, clarifying.
Within the age of AI coding brokers, the standalone app has an issue.
When constructing an app takes days as a substitute of years, the apps themselves are price much less. We’re already dwelling an early model of this. My firm makes use of a hiring platform for recruiting, however we barely contact its frontend anymore. We constructed our personal interface on prime of its API — one which higher matches how our group really works. The platform has successfully turn into a glorified database, and I don’t suppose we’ll renew.
That’s the shift founders want to concentrate to. Worth in software program is transferring away from the interface and towards the information beneath it. An app frontend is more and more a legal responsibility: opinionated in methods customers didn’t select, maintained on another person’s timeline, and constructed for a median use case fairly than yours. The founders who don’t see this coming will spend years constructing an interface that their clients will finally substitute themselves.
What fills the hole is one thing I consider because the “meta-app”: apps that construct different apps on the fly, completely tailor-made to your fast want. You inform it what you need, and it figures out the remainder. AI coding instruments like Claude Code and Codex are an early model of this. They’re getting a variety of consideration for serving to builders construct software program sooner, however their actual energy is larger than that. It’s the flexibility to go instantly from intent to consequence, for any activity, with out an app within the center.
My group used our personal AI coding platform, Warp, to construct the Sheets clone, and groups throughout the group are utilizing it to spin up customized instruments that substitute the apps they used to depend on. Our artistic director, who doesn’t know write code, makes use of it as his start line for nearly something. He not too long ago migrated our outdated web site off its CMS simply by describing what he needed. He didn’t want an app to do this — simply one thing that understood what he was making an attempt to do.
The counter-argument I hear most frequently is that the meta-app nonetheless has to connect with current techniques in some way — by means of APIs, integrations, and connectors. That’s true right now, however these layers exist as a result of knowledge is at the moment siloed throughout legacy apps. They’re a transitional resolution. The businesses that construction their knowledge for brokers now, fairly than locking it inside app interfaces, shall be higher positioned as that transition accelerates. At Warp, we retailer firm knowledge in easy information and databases fairly than in SaaS merchandise, as a result of we don’t need to spend years doing integration work simply to make our personal info out there to the brokers that want it.
For any firm nonetheless closely invested within the present app mannequin, there’s one other downside price taking note of. Many SaaS firms will reply to this shift by making an attempt to make it tougher to get your knowledge out, as a result of the information they maintain is more and more their solely defensible asset. Some will attempt to lock clients into their very own agentic options, retaining something you construct depending on their techniques. Which may purchase them time, nevertheless it received’t change the place issues are headed.
Apps received’t disappear in a single day. The transition shall be uneven, and a few classes will maintain longer than others. Stripe isn’t going wherever. Neither is Spotify. Corporations with genuinely advanced enterprise logic, or with distinctive knowledge at their core, aren’t going wherever both. A frontend sitting on prime of a database, even a classy one, is a unique story. The window for constructing a defensible software program enterprise round that type of interface is closing sooner than most individuals inside these companies need to consider.
I understand this may sound threatening in case you’ve spent years, as I’ve, pondering rigorously about construct nice apps. However I believe it’s really probably the most thrilling factor occurring in software program proper now. We’re transferring towards a world the place anybody can construct what they want, simply by describing what they need. We’ll all be software program builders, even when none of us are writing code, and software program will lastly serve us, fairly than the opposite approach round.
The opinions expressed in Fortune.com commentary items are solely the views of their authors and don’t essentially replicate the opinions and beliefs of Fortune.