Google Chrome Tests Gemini-Powered 'Skills' for Automated Browser Tasks

Google is advancing its vision for AI-powered browsing with experimental 'Skills' for Gemini in Chrome. Discovered in Chrome's internal builds, this feature transforms the AI assistant from a helpful companion into a proactive agent capable of executing multi-step tasks autonomously.

Currently, Gemini operates as a contextual helper within Chrome. Users can summon it to explain webpage content, summarize articles, or compare information across tabs—like compiling flight, hotel, and activity details into a unified travel plan. The upcoming Skills system, accessible via a dedicated chrome://skills interface, enables users to create custom automations by combining:

1. Task names (e.g., "Expense tracker")
2. Natural language instructions ("Scan Gmail for receipts, extract vendor/amount, populate spreadsheet")

This framework suggests Google's move toward agentic browsing, where Gemini could handle repetitive workflows without manual intervention. Internal documentation indicates Skills will eventually integrate tightly with Google services—imagine asking:

"Find YouTube videos from my history about 3D modeling that are under 10 minutes"

without switching tabs or apps.

While still in early testing with no public release timeline, the infrastructure aligns with Google's confirmation that Gemini would evolve into a proactive Chrome agent. The Skills approach offers more flexibility than predefined features, potentially allowing users to build personalized automations like research compilation or price monitoring.

Privacy implications remain unaddressed, particularly regarding Skills accessing Gmail or browsing history. As development continues, balancing automation power with user control will be crucial for adoption. This represents a significant step beyond today's reactive AI helpers toward truly assistive browsing experiences.