The exceptionally fast evolution of AI tooling has left many of us in the Android Dev world in a state of confusion: with so many new tools popping up so quickly, staying up-to-date and knowing what to use, when, and how has become a challenge. In this session, we’ll talk about the modern AI tooling landscape for Android, and how these seemingly fragmented parts can operate as a cohesive, tooling ecosystem. You can also mix & match to create your own optimal AI developer productivity stack - from the deeply integrated Agent Mode in Android Studio, to agent-first workspaces like Antigravity; Android CLI which acts as the glue between surfaces, enabling headless execution and unlocking IDE functionality for any AI agent, including Claude and Codex; Android skills to help ground and orchestrate LLMs with freshest domain knowledge; how to rely on local, open-weights models, like Gemma 4, for repetitive, daily tasks without overloading context windows. Finally, we’ll dive under the hood of Android Bench to show how we objectively evaluate LLMs and regression-test them against real-world Android development problems. We hope this session will help you truly understand and choose more wisely the tooling for your daily workflows.