On Wednesday, Google unveiled Androidify, a new platform available as both a website and a standalone Android app. This innovative tool harnesses some of Google’s most advanced AI models, including the recently introduced “Nano-Banana” image generator from Gemini (Gemini 2.5 Flash Image), to transform selfies into personalized Android bots. Notably, the service does not replicate users’ faces on the bots; instead, it captures general features and accessories, allowing users to craft their ideal Android avatars. Additionally, Androidify accommodates text prompts, enabling users to generate bots based on specific design instructions.
Androidify stands out as the more whimsical aspect of Google’s broader Android-related announcements made on the same day. This early September Pixel Drop also introduced Material 3 Expressive design enhancements for a wider range of Pixel phones, alongside new functionalities for the Pixel Buds Pro 2 and Pixel Watch. Furthermore, Google rolled out several updates for non-Pixel devices, including AI writing tools in Gboard, enhancements to Emoji Kitchen, a redesign of Quick Share, and improved audio sharing capabilities.
How Androidify uses Nano-Banana to create Android bots
In a comprehensive blog post on the Android Developers blog, Google elaborated on the AI technologies powering the Androidify app. The process begins with users uploading a selfie on Androidify.com or through the mobile app, or alternatively, providing a text prompt that outlines their desired Android bot design. Following this, Androidify employs Gemini 2.5 Flash to validate the uploaded photo, ensuring it features a clear, focused individual and adheres to safety standards before proceeding with further processing.
The AI model then generates a caption for the image, which is transmitted in JSON format to the subsequent AI model. Google has tailored a version of its Imagen 3 image generation tool specifically for creating custom Android bot avatars. This model interprets the JSON-enriched prompt to produce a unique Android bot that aligns with user specifications. Additionally, Gemini 2.5 Flash is integrated into the app, allowing users to generate random descriptions for clothing and hairstyles through a “Help me write” feature.
The Nano-Banana model, which gained attention in August prior to its official introduction as Gemini 2.5 Flash Image, is utilized within Androidify to craft backgrounds for the custom Android bots, resulting in distinctive AI-generated scenes featuring the avatars. The Androidify app also includes a Sticker mode, enabling users to eliminate backgrounds and create stickers based on their custom bots.
Moreover, Google incorporates Veo 3, its premier video-generation model, to allow users to animate their Androidify creations. Throughout September, Android users can produce 8-second videos from their Androidify images every Friday. A demonstration video is available, showcasing the capabilities of the Androidify app.
