Build intelligent Android apps with Google's AI
summary
Claims concrete outcomes: Gemma 4 → Gemini Nano runs on-device via AI Core and benchmarks show up to 4× faster, 60% less battery for reasoning and image tasks. Available tools: ML Kit GenAI APIs for turnkey summarization, prompt and speech (50 MB multilingual Nano), Light RTLM for custom models with device benchmarking on Google-hosted phones. Hybrid cloud via Firebase AI Logic, Gemini Live low-latency audio, function calling, ADK + AGUI enable multi-agent orchestration.
AI-written summary · about
chapters (18)
- 0:00Building AI Android apps
- 1:26Gemma 4 and Gemini Nano 4
- 4:37AI pillars on Android
- 5:38In-app AI categories
- 6:18On-device inference benefits
- 7:42ML Kit GenAI APIs
- 10:52Feature vs prompt APIs
- 14:21On-device audio with ML Kit
- 16:28Custom models with Light RTLM
- 19:58Hybrid cloud-device inference
- 22:08Cloud inference overview
- 24:36Firebase AI Logic
- 27:49Gemini Live API
- 30:38Function calling for agents
- 33:29Scaling multi-agent systems
- 38:23AGUI protocol
- 39:07AI solutions summary
- 40:10Resources and closing
related talks
30mGenAI on Android with Gemini Nano and Firebase
39mAdd Generative AI capabilities to your Android app
21mAdd Generative AI capabilities to your Android app
10mAI in your pocket: Building intelligent Android apps
41mGen AI for Android Developers
20mGemini Nano on device
20mWhat's new in Machine Learning Keynote
23m2025: On‑Device AI Goes Mainstream on Android