AI app generation: from idea to app tsunami
AI app generation is the process by which artificial intelligence models and code-writing agents transform high-level ideas or rough prototypes into functioning applications that can be packaged, validated, and distributed through app stores or internal channels without requiring humans to write most of the underlying code. The App Store is evolving at a breakneck pace not seen since its earliest days, driven by a tsunami of apps developed with the help of AI agents that only really took hold late last year. We saw the first glimmers of this wave early in 2025, but it is now a flood: supply is dramatically up, demand is flat, and quality is uneven, depending on where you look. In practical terms, AI-powered app creation has made it far easier to build a web or native app yourself, thanks to frontier labs’ model improvements and a growing ecosystem of app-building tools from those labs and third parties.
Software factories AI: the new production line for code
AI-powered software factories are the industrial counterpart to this app explosion. Imagine submitting a vibe-coded initial version of an app to a service that mass-produces your solution in a repeatable, automated way for wide distribution. That is no longer a thought experiment; today’s software factory is built on AI models, with coding capabilities of foundation models and associated agents reviving the concept and breaking the old bottleneck where organizations needed to hire more engineers to write code. With agentic coding, teams are less constrained on the coding side, and developer craftsmanship shifts from typing lines of code to exercising judgment about what should be built and shipped. Over the past 18 months, leading AI companies such as Anthropic, Cognition, Cursor, Factory, Google, Github, OpenAI, and Ramp have converged on the same “machine” for automating the software development lifecycle, and that convergence is what is now spilling over into public app stores.
Utility apps: the first casualties of automated software development
If you want to see the impact of automated software development, look at utility apps. Many utilities are little more than UI wrappers around command line tools, the classic example being a friendly interface around ffmpeg so users do not have to open Terminal to convert a video or audio file. Their relative simplicity and narrow scope make them a natural fit for AI agents, which is why the App Store and reviewers’ inboxes are deluged with Mac menu bar utilities and single-screen iPhone and iPad apps. According to one analysis, “app supply is outstripping demand by orders of magnitude, which will drive down prices,” based on data reported by a major newspaper. Utilities have even pushed their category into the top five charts in app analytics data. On the one hand, more utilities give users better odds of finding a perfect fit; on the other, this flood looks unsustainable, with utility downloads expected to shrink in the coming months as agents make it trivial for power users to build their own tools.
From customer support ideas to factory-built apps
What makes this shift so disruptive is who can now participate. AI software factories let someone in a customer support role turn their daily insight into a working application: they submit an idea based on feedback they hear, the factory produces a first iteration, and then product, engineering, and IT review and refine it before sending it back to the factory for scaled production. In other words, developers can feed vibe-coded prototypes or natural language descriptions into an automated pipeline that takes care of much of the heavy lifting. For ordinary users, this can be a win. Most people do not want to tinker with terminals or scripting when they can download an app that is easier to use. More choice means you are more likely to find the app that fits your needs. But when the barrier to creation falls this low, the line between “user” and “developer” blurs, especially in the narrow world of utilities where power users can now roll their own tools.
The next phase: when validation becomes the bottleneck
The App Store’s current growth mirrors its earliest gold rush, but automation has turned the dial from fast to frantic: download numbers across the store have only grown 2–3% this year and last, yet AI-assisted development has pushed supply far ahead of demand by orders of magnitude. That imbalance will not resolve itself kindly. Utility app markets are expected to contract as agents empower power users to build their own tools, and as prices slide in a crowded field. Meanwhile, software factories have their own looming breakpoint. They help address the challenge of validating and maintaining hundreds of thousands of lines of code and the daily cadence of releases and updates, but they also push us toward a future where, in the words of one industry leader, “we will get to the point where as humans we can’t validate what’s working and what’s not really working anymore”. The real question is not whether AI-powered app creation will remake software; it is who will control the factories, and who will be left to sort through the flood.
