New Technology / Ai Agents
Google Gemini Skills in Chrome
Track AI agents, autonomous workflows, agentic software tools and real-world adoption signals across the next wave of AI products.
Source material: Google New Gemini Skillz Turn Chrome Into an AI Beast
Key insights
- Googles introduction of Skills in Chrome allows users to save prompts as reusable workflows, reducing repetitive typing across multiple pages
- The new feature enables skills to function across multiple tabs, enhancing the browsers capability as a powerful retrieval system
- Googles built-in library provides users with curated prompts for various tasks, making advanced prompt management accessible to all
- Safety measures require user confirmation for high-impact actions initiated by skills, addressing concerns about automated workflows executing irreversible actions
- The launch of Skills in Chrome is a key step towards developing browser-level AI agents capable of managing persistent workflows and complex reasoning tasks
- Google is also testing a new agent tab in Gemini Enterprise, signaling a move towards a more comprehensive task execution system for better collaboration
Perspectives
Analysis of Google Gemini Skills and its implications.
Proponents of Google Gemini Skills
- Introduces Skills in Chrome for reusable workflows
- Enhances browser functionality as a retrieval system
- Allows users to save prompts and trigger them easily
- Implements safety measures for high-impact actions
- Facilitates multi-tab reasoning and persistent workflows
Critics of Google Gemini Skills
- Overlooks potential user confusion regarding prompt management
- Assumes users will benefit from reusable workflows without adequate training
- Raises concerns about the complexity of new features
Neutral / Shared
- Integrates features like Require Human Review for task execution
- Develops a built-in library of pre-made skills for user convenience
Metrics
other
Skills in Chrome
introduction of a new feature
This marks a significant advancement in browser capabilities.
Google just rolled out something called Skills in Chrome
other
available on Mac, Windows, and Chrome OS
platform availability
Wider accessibility increases potential user adoption.
It's available on Mac, Windows, and Chrome OS.
other
confirmation gates for high impact actions
safety measures
This addresses user concerns about automated actions.
Google added confirmation gates for high impact actions.
success_rate
86%
Gemini Robotics ER 1.6 performance
A higher success rate indicates improved reliability in robotic tasks.
Gemini Robotics ER 1.6 reaches 86%.
success_rate
23%
Gemini Robotics ER 1.5 performance
This shows the significant improvement in the latest version.
Gemini Robotics ER 1.5 had a 23% success rate.
success_rate
67%
Gemini 3.0 flash performance
Demonstrates the progression in success rates across versions.
Gemini 3.0 flash reached 67%.
success_rate
93%
Gemini Robotics ER's performance in interpreting gauges
A high success rate indicates improved reliability in robotic applications.
And with agentic vision enabled, it hits 93%.
evidence_rate
92.4%
Project management conversation level information rates
High evidence rates suggest effective assessment of project management skills.
For project management, conversation level information rates reached 92.4%.
Key entities
Timeline highlights
00:00–05:00
Google has introduced Skills in Chrome, allowing users to save prompts as reusable workflows across multiple tabs. This feature enhances the browser's functionality as a retrieval system and includes safety measures for high-impact actions.
- Googles introduction of Skills in Chrome allows users to save prompts as reusable workflows, reducing repetitive typing across multiple pages
- The new feature enables skills to function across multiple tabs, enhancing the browsers capability as a powerful retrieval system
- Googles built-in library provides users with curated prompts for various tasks, making advanced prompt management accessible to all
- Safety measures require user confirmation for high-impact actions initiated by skills, addressing concerns about automated workflows executing irreversible actions
- The launch of Skills in Chrome is a key step towards developing browser-level AI agents capable of managing persistent workflows and complex reasoning tasks
- Google is also testing a new agent tab in Gemini Enterprise, signaling a move towards a more comprehensive task execution system for better collaboration
05:00–10:00
Google is enhancing its Gemini system to function as a comprehensive workspace for multi-step workflows, indicating a shift towards a desktop agent capable of executing complex tasks. The introduction of features like 'Require Human Review' and the visual layer in NotebookLM signifies a commitment to improving user control and data presentation.
- Google is transforming its Gemini system into a versatile workspace for multi-step workflows, indicating progress towards a fully functional desktop agent capable of handling complex tasks
- The addition of a Require Human Review toggle indicates that Google is preparing agents to perform real actions, which enhances user control over automated processes
- NotebookLM is being upgraded with a visual layer called Canvas, enabling users to create structured and interactive data presentations, moving beyond traditional document analysis
- The new connectors feature in NotebookLM will allow it to access data from external services, enhancing its role as a central research tool and improving data accessibility
- Gemini Robotics ER 1.6 features significant enhancements in spatial reasoning and success detection, which are crucial for improving the reliability of robotic systems in real-world scenarios
- The capability of Gemini Robotics ER 1.6 to interpret analog gauges and digital displays marks a major advancement in robotic perception, significantly boosting operational efficiency
10:00–15:00
Gemini Robotics ER has achieved a 93% success rate in interpreting real-world gauges, indicating a significant advancement in robotic capabilities. Google Research's Vantage system evaluates human skills like collaboration and creativity, demonstrating superior performance in assessing project management and conflict resolution skills compared to independent agents.
- Gemini Robotics ER has achieved a 93% success rate in interpreting real-world gauges, marking a significant advancement in robotic capabilities for complex environments
- Google Researchs Vantage system evaluates human skills like collaboration and creativity, filling gaps in traditional assessment methods by using a single executive LLM to guide interactions
- In trials, Vantage demonstrated superior performance in assessing project management and conflict resolution skills compared to independent agents, achieving high evidence rates
- The accuracy of Vantages AI assessments closely aligns with human evaluators, indicating its potential as a reliable tool for evaluating complex human interactions in educational contexts
- Vantage can simulate participants with varying skill levels, which may streamline testing processes and reduce reliance on expensive human studies
- Results from Vantage are visualized as a skills map, offering actionable insights into competency levels tied to specific conversation segments