Google Launches ADK for Kotlin on Android to Build AI Agents
ADK for Kotlin

Google has announced Agent Development Kit (ADK) support for Kotlin and Android, allowing developers to build AI-powered agents directly inside Android applications using native Kotlin workflows. The announcement was made through the official Google Developers Blog during Google I/O 2026.  

According to Google, ADK for Android is an open-source framework designed for building AI agents that can run locally on Android devices, inside hosted services, or across hybrid environments. The framework supports both Kotlin and Java development.  

Google says developers can now create sophisticated AI agents capable of:

  • Tool calling

  • Workflow orchestration

  • Multi-agent interactions

  • On-device AI execution

  • Cloud-connected reasoning

  • Long-running task automation

The company is positioning ADK as part of its broader “agentic AI” strategy announced across Gemini 3.5 Flash, Antigravity 2.0, Managed Agents, and the Gemini API ecosystem during I/O 2026.  

One of the biggest advantages for Android developers is native Kotlin integration. Google says developers can build AI agents using familiar Android tooling and architecture patterns instead of relying entirely on external AI platforms. The framework also integrates with Android AI capabilities and Gemini-powered services.  

Google highlighted that ADK supports scalable multi-agent architectures where specialized agents can coordinate tasks together inside mobile applications. This reflects the growing industry trend toward autonomous AI systems instead of simple chatbot experiences.  

The launch also aligns with Google’s larger push into AI-native Android development. During I/O 2026, the company additionally announced:

  • AI-powered Android app generation in Google AI Studio

  • Android CLI for AI coding agents

  • Antigravity 2.0 agent workflows

  • Managed Agents inside Gemini API

  • Gemini-powered Android integrations

Google says ADK for Android supports both local and cloud execution models, giving developers flexibility for privacy-sensitive workloads, offline execution, and scalable backend orchestration.  

The framework is available now through Google’s Android AI developer resources.