A lightweight Kotlin library that lets Android developers integrate Google’s Gemini AI into their apps with ease. Supports text, image, and multi-modal inputs, persistent chat history, and agentic task detection with JSON-based structured responses for automation.

Prexo - AI

Integrate Google Gemini AI into your Android apps — now with Agentic Task Detection

PrexoAI is a lightweight Kotlin library that makes it effortless to bring Gemini AI’s capabilities to your Android projects. With support for text, image, and multi-modal inputs, persistent conversation history, and smart agentic task detection, you can go beyond chat — automate actions, parse structured JSON commands, and build futuristic AI-powered experiences.

Features

• Simple Gemini integration with minimal setup.
• Multi-modal support for text, images, and combined inputs.
• Agentic task detection returning structured JSON.
• Persistent memory for contextual conversations.
• Configurable model, safety settings, and system prompts.
• Quick and lightweight.

Installation

Add JitPack to your settings.gradle.kts:

dependencyResolutionManagement {
    repositoriesMode.set(RepositoriesMode.FAIL_ON_PROJECT_REPOS)
    repositories {
        google()
        mavenCentral()
        maven { url = uri("https://jitpack.io") }
    }
}

And include the dependency in your build.gradle.kts:

dependencies {
    implementation("com.github.binarybeam:Prexo-Ai:1.3.0")
}

Quick Start

val ai = Ai()
ai.updateModel(apiKey = "YOUR_API_KEY", agentMode = true)

ai.sendMessage("Turn on the flashlight", object : AiCallback {
    override fun onNormalResponse(message: String) { println("AI: $message") }
    override fun onAgenticResponse(message: String, jsonObject: List<JSONObject>) {
        println("Agentic Task: $message")
        println("Actions: $jsonObject")
    }
    override fun onError(error: String) { println("Error: $error") }
})

Agentic Mode

When enabled, responses for actionable tasks are returned in JSON format only.

{
  "id": "wifi",
  "state": "on",
}

Configuration

ai.updateModel(
    apiKey = "YOUR_API_KEY",
    modelName = "gemini-2.5-flash",