Google Lets Anyone Build Android Apps From a Browser

Google announced at its annual developer conference, Google I/O, that its free web-based tool Google AI Studio can now build complete Android apps from a plain-language description. You open a browser, type what you want the app to do, and the AI writes and assembles it. No software to download, no technical setup required. The apps are real, not just demos. They run natively on Android phones and can connect to hardware features like GPS for location tracking, Bluetooth for connecting to devices, and NFC for tap-based interactions. A preview of the app runs directly inside the browser so you can test it before putting it on your phone. The current limitation is worth being clear about: right now, apps built this way are for personal or internal use only. Publishing to the general Play Store for public download is not yet available, though Google says broader publishing, including sharing with friends and family, is on the roadmap. For those who want to go further, the tool handles the technical packaging automatically. It can upload your app directly to Google Play's testing track so you can install and update it on your phone as you refine it. Developers who want to hand the project to a technical team can export the full project as a file or push it directly to GitHub, the platform where developers store and share code. The broader context matters here. The market for AI-powered app building has exploded. Platforms like Lovable, Cursor, and Replit have collectively attracted tens of millions of users and billions in investment over the past 18 months. Roughly 63% of people using these tools are not developers at all. Google is entering this market with a significant advantage: it owns Android, it owns the Play Store, and its tool is free. What Google is really doing with this launch is two things at once. It is pulling non-technical builders into its own orbit rather than letting them build on third-party platforms. And it is extending its AI assistant, Gemini, into app discovery: in the coming weeks, Gemini will begin recommending Play Store apps directly inside its chat interface, across Android phones and the web. Later this year, that same discovery layer will cover over 450,000 movies and TV shows, and live sports streaming links. The discovery change is the part most business operators should pay attention to. If you have an app, or if you build one, the route to users is no longer just the Play Store search bar. It increasingly runs through conversational AI, specifically through people asking Gemini what app they should use. That is a meaningful shift in how software gets found. The security question is real and should not be ignored. Across the broader AI app building market, studies have found that AI-generated code carries meaningfully higher rates of security flaws than code written by experienced developers. For personal utilities and internal prototypes, this is a manageable risk. For anything touching customer data, payments, or regulated information, it warrants proper review before deployment. The simplest way to think about this: Google has just made the gap between having an idea for an app and having a working version of it very small. The gap between that working version and something you would put in front of paying customers or regulated users is still real, but the cost of reaching the first milestone has dropped dramatically.