We’re in an essential second the place AI modifications all the things, from how we work to the expectations that customers have to your apps, and our objective on Android is to rework this AI evolution into alternatives for you and your customers. At present in our Fall episode of The Android Presentwe unpacked a bunch of latest updates in direction of delivering the very best return on funding in constructing for the Android platform. From new agentic experiences for Gemini in Android Studio to a model new on-device AI API to the primary Android XR machine, there’s a lot to cowl – let’s dive in!
Construct your individual customized Gen AI options with the brand new Immediate API
On Android, we provide AI fashions on-device, or within the cloud. At present, we’re excited to now offer you full flexibility to form the output of the Gemini Nano mannequin by passing in any immediate you’ll be able to think about with the brand new Immediate API, now in Alpha. For flagship Android gadgets, Gemini Nano enables you to construct environment friendly on-device choices the place the customers’ knowledge by no means leaves their machine. At I/O this Might, we launched our on-device GenAI APIs utilizing the Gemini Nano mannequinmaking frequent duties simpler with easy APIs for duties like summarization, proofreading and picture description. Kakao used the Immediate API to rework their parcel supply service, changing a gradual, guide course of the place customers needed to copy and paste particulars right into a type into only a easy message requesting a supply, and the API routinely extracts all the mandatory data. This single characteristic lowered order completion time by 24% and boosted new person conversion by an unimaginable 45%.
Faucet into Nano Banana and Imagen utilizing the Firebase SDK
If you need to add cutting-edge capabilities throughout the whole fleet of Android gadgets, our cloud-based AI options with Firebase AI Logic are an incredible match. The joy for fashions like Gemini 2.5 Flash Picture (a.okay.a. Nano Banana) and Picture have been unimaginable; now your customers can now generate and edit pictures utilizing Nano Banana, after which for finer management, like deciding on and reworking particular components of a picture, customers can use the brand new mask-based enhancing characteristic that leverages the Imagen mannequin. See our weblog publish to be taught extra. And past picture technology, it’s also possible to use Gemini multimodal capabilities to course of textual content, audio and picture enter. RedBus, for instance, revolutionized their person opinions utilizing Gemini Flash through Firebase AI Logic to make giving suggestions simpler, extra inclusive, and dependable. The previous drawback? Quick, low-quality textual content opinions. The brand new resolution? Customers can now go away opinions utilizing voice enter of their native languages. From the audio Gemini Flash is then producing a structured textual content response enabling longer, richer and extra dependable person opinions. It is a win for everybody: vacationers, operators, and builders!
Serving to you be extra productive, with agentic experiences in Android Studio
Serving to you be extra productive is our objective with Gemini in Android Studioand why we’re infusing AI throughout our tooling. Builders like Pocket FM have seen a powerful growth time financial savings of fifty%. With the latest launch of Agent Modeyou’ll be able to describe a posh objective in pure language and (along with your permission), the agent plans and executes modifications on a number of recordsdata throughout your venture. The agent’s solutions are actually grounded in probably the most trendy growth practices, and may even cross-reference our newest documentation in actual time. We demoed new agentic experiences corresponding to updates to Agent Mode, the flexibility to improve APIs in your behalf, the brand new venture assistant, and we introduced you’ll be capable to convey any LLM of your option to energy the AI performance inside Android Studio, supplying you with extra flexibility and selection on the way you incorporate AI into your workflow. And for the latest steady options corresponding to Again Up and Syncbe certain to obtain the most recent steady model of Android Studio.
Elevating AI-assisted Android growth, and enhancing LLMs with an Android benchmark
Our objective is to make it simpler for Android builders to construct nice experiences. With extra code being written by AI, builders have been asking for fashions that know extra about Android growth. We need to assist builders be extra productive, and that’s why we’re constructing a brand new process set for LLMs towards a spread of frequent Android growth areas. The objective is to supply LLM makers with a benchmark, a north star of top of the range Android growth, so Android builders have a spread of useful fashions to decide on for AI help.
