generative-ai-android

Google AI SDK for Android: Project Overview

The Google AI Android SDK is an accessible tool designed for Android developers who wish to integrate the sophisticated capabilities of the Gemini API into their applications. This API, crafted by the brilliant minds at Google DeepMind, offers access to Gemini models, which are inherently multimodal. This means they are adept at smoothly processing and reasoning across different types of input—namely text, images, and code.

Important Considerations

When working with the Google AI SDK for Android, one should note that it is primarily intended for prototyping. Developers are advised against embedding their API keys directly into their Android applications or fetching them during runtime, as this could expose the keys to malicious entities. For production environments, using a backend SDK to interact with the Google AI Gemini API is strongly recommended. Moreover, if there are requirements for on-device model usage, such as Gemini Nano, developers should explore the Google AI Edge SDK for Android.

Getting Started with the Gemini API

This project includes a sample app that serves as a demonstration of how the SDK can effectively utilize the Gemini model for various applications. Developers keen to try out this sample app can do so by importing the project directly from Android Studio or by following these steps:

1. Visit Google AI Studio and log in with a Google account.
2. Create an API key via the API key creation page; be aware that the free tier is unavailable in Europe.
3. Clone the repository with: git clone https://github.com/google/generative-ai-android.
4. Navigate to the generativeai-android-sample folder, open it, and build the sample app.
5. Insert your API key into the apiKey field in the local.properties file.
6. Run the app to observe its functionalities.
7. For comprehensive guidance, refer to the Android SDK tutorial.

Usage Example

To begin integrating the Gemini API into an Android project, developers need to:

1. Add the dependency:

   implementation("com.google.ai.client.generativeai:generativeai:<version>")
   

2. Initialize the model using the following code snippet:

   val generativeModel = GenerativeModel(
       modelName = "gemini-1.5-pro-latest",
       apiKey = BuildConfig.apiKey
   )
   

3. Execute a prompt as demonstrated below:

   val cookieImage: Bitmap = // ...
   val inputContent = content() {
     image(cookieImage)
     text("Does this look store-bought or homemade?")
   }
   
   val response = generativeModel.generateContent(inputContent)
   print(response.text)
   

For extended instructions, including additional use cases and features such as streaming, token management, and response customization, refer to the quickstart guide.

Documentation and Contribution

For those seeking in-depth information, the Gemini API Cookbook and Google's official documentation are invaluable resources.

Developers interested in contributing to the Google AI client SDK for Android can find relevant details in the contribution guidelines.

Licensing

The project repository content is available under the Apache License, version 2.0, ensuring open access and modification rights.

In summary, the Google AI SDK for Android empowers developers with advanced AI capabilities, particularly in processing diverse data forms. This project offers substantial resources and guidance to support developers in efficiently leveraging the Gemini API.