The latest AI news we announced in March 2026
Key Points
- Gemini 3.1 Flash‑Lite: low‑latency, cost‑efficient model
- Search Live expanded globally with voice/camera dialogue
- Lyria 3 Pro and Gemini API available in public preview
Summary
In March 2026 Google shipped a broad set of AI updates focused on faster models, developer tooling, and deeper integration of Gemini across products. Releases include new low-latency Gemini models for real-time apps, expanded conversational Search Live, workspace-level AI in Docs/Sheets/Drive, developer previews for creative models, and tools to migrate user context into Gemini.
Key Points
- Models and latency: Released Gemini 3.1 Flash‑Lite (fast, cost‑efficient for high-throughput workloads) and Gemini 3.1 Flash Live (low‑lag audio model for conversational experiences).
- Developer APIs and previews: Lyria 3 Pro and Lyria/Lyria 3 are in public preview via the Gemini API and Google AI Studio; these enable programmatic music generation up to 3 minutes with fine-grained controls.
- Google AI Studio improvements: New Antigravity coding agent and Build mode to convert prompts into production‑ready apps, manage project context, store API keys securely, and connect databases/services.
- Search and UI integration: Search Live expanded to 200+ countries with back‑and‑forth voice/camera interaction and Canvas for multi‑step project workflows; helpful for heads‑up, real‑time tooling.
- Workspace and data synthesis: Gemini features in Docs/Sheets/Slides/Drive for AI Ultra/Pro subscribers, with secure synthesis across files and state‑of‑the‑art performance in Sheets for complex analysis.
- Migration and personalization: New memory and chat history import tools to port preferences and chats into Gemini; Personal Intelligence expanded to more surfaces with user-controlled connections to Gmail, Photos, etc.
- Maps and device features: Ask Maps (conversational queries), Immersive Navigation with real‑world imagery, Pixel Drop device features (Circle to Search breakdowns, Magic Cue recommendations, Pixel Watch updates), and expanded live translation on iOS/Android headphones.
- Health and partnerships: $10M funding for clinician education initiatives, Fitbit personal health coach expansion (medical records connectivity, cycle/mental/ nutrition features) and rural health partnerships.
Engineering impact and next steps
- Consider Flash‑Lite for latency‑sensitive, cost‑constrained real‑time endpoints; evaluate Flash Live for voice interfaces.
- Explore Gemini API and Google AI Studio previews (Lyria and Antigravity agent) for prototyping creative features and faster app iteration.
- Update integrations where personalization requires user data: ensure consent flows and settings for Personal Intelligence imports and chat migration.
- Test Search Live and Ask Maps integration points for conversational UIs and multi‑step workflows in mobile and in‑car contexts.
Availability and privacy notes
- Search Live, Flash Live, and related features rolled out in 200+ countries where AI Mode is supported; Canvas and some features are region/language limited (e.g., Canvas in the U.S. in English).
- Personal Intelligence and file synthesis are opt‑in; users control connected data sources in settings.