Google AI: Release Notes Gemini in Chrome: Your agentic browsing assistant
26 snips
Mar 12, 2026 They explore embedding an agentic Gemini into Chrome to perform web tasks like research and shopping. The conversation covers Nano Banana for in-context media edits and demos of auto browse handling multi-step workflows. They tackle tab and context overload, side panel multitasking, safety guardrails, standards for agentic actions, on-device models, and scaling design for billions of users.
AI Snips
Chapters
Transcript
Episode notes
Chrome As A Contextual Platform
- Chrome functions like an OS-on-top that carries personal context (history, autofill, passwords) across devices for longer multi‑device journeys.
- That context enables Gemini to help with long-lived tasks like shopping, summarizing lectures, and synthesizing across tabs.
Nano Banana Edits Images Inline
- The team integrated Nano Banana so users can edit images inline (e.g., change room colors) without downloading/uploading to other apps.
- Rick demoed changing wall color and adding cushions directly on a Redfin image.
Use History Recall To Close Tabs
- Try Chrome's history recall instead of hoarding tabs to avoid loss aversion and confidently close tabs.
- Ask Gemini to retrieve recently closed items like the three Italian restaurants you searched for last week.