<p>Google has updated AI Studio with a set of features that push it well beyond a frontend prototyping sandbox. The headline additions are native Firebase integration, an autonomous coding agent called Antigravity, and a share button for multiplayer testing before you ever touch a deployment pipeline.</p>
<h2>What’s actually new</h2>
<p><strong>Firebase integration</strong> is probably the biggest shift. Previously, wiring up a database or authentication meant leaving AI Studio and doing it yourself. Now a single prompt can provision a Firestore database and set up Google Auth. The system detects when your app needs persistent storage and handles it automatically.</p>
<p><strong>The Antigravity agent</strong> works in a loop: it writes code, runs it, checks the output, and corrects errors in the background without you asking. It also installs missing npm packages on its own after reading your project structure. Crucially, it maintains context across your entire project and across sessions - close the tab, come back the next day, and it picks up where it left off.</p>
<p><strong>Secrets management</strong> lets you connect third-party APIs (payments via Stripe, maps, etc.) without hardcoding keys into your project.</p>
<p><strong>Deployment</strong> goes directly to Google Cloud Run with one button.</p>
<h2>A concrete example</h2>
<p>One demo built during the announcement was a typing test app with a real-time leaderboard. Multiple users could sign in with Google Auth, scores were written to Firestore, and the leaderboard updated live across sessions. The whole thing was assembled through prompts, with Antigravity handling the error correction loop in the background.</p>
<p>Other demos included multiplayer laser tag, 3D physics games, and apps pulling live Google Maps data - all reportedly built from single prompts.</p>
<h2>How it fits into the broader landscape</h2>
<p>This puts Google in direct competition with Claude Code and OpenAI Codex, but with a meaningful advantage: Google already owns the infrastructure stack. Calendar, Gmail, Docs, Maps, and now the IDE are all under one roof. Whether that’s reassuring or unsettling probably depends on how much of your stack you already run on Google.</p>
<p>The target audience is wide - rapid prototyping for developers, MVP builds for startups, and in theory, app creation for people without coding experience. Build Mode handles frontend UI, backend logic, and APIs from natural language descriptions.</p>
<p>One thing worth noting: Apple reportedly moved to block vibe coding apps from updating in the App Store the same week Google shipped this. Probably coincidence. Probably.</p>


Collections

Google AI Studio has been upgraded with full-stack app building capabilities. Key additions include native Firebase integration (Firestore database and Google Auth provisioning via prompts), an autonomous coding agent called Antigravity that writes, runs, and corrects code in a loop while maintaining cross-session context, secrets management for third-party APIs, and one-click deployment to Google Cloud Run. The platform now targets developers for rapid prototyping, startups building MVPs, and non-coders building apps through natural language. This positions Google in direct competition with Claude Code and OpenAI Codex, with the advantage of owning the full infrastructure stack including Calendar, Gmail, Maps, and cloud hosting.

Google AI Studio adds full-stack app building with Firebase, auth, and one-click deploy