Overview
- On Thursday, Google Cloud launched the Gemini agent, a versatile AI designed for workplace tasks, operating seamlessly across Gmail, Docs, Sheets, Slides, Drive, Chat, and Calendar.
- Managers can establish "coworker agents" featuring their own Workspace email, calendar, Drive storage, and a listing in the company directory, with all actions logged separately from human users.
- Google has not yet disclosed pricing or a release date; however, versions tailored for financial services and legal sectors are currently in preview, with government, healthcare, and retail versions expected soon.
Google Cloud has unveiled its Gemini Agent, a comprehensive AI tool for the workplace that can autonomously tackle tasks and report back with the results. During the Gemini at Work event, CEO Thomas Kurian emphasized that "work now starts in the prompt window."
This Gemini Agent integrates within various platforms, including Gmail, Docs, Sheets, Slides, Drive, Chat, and Calendar, and even extends its functionality to Microsoft 365 and Slack. Google asserts that the AI continues to operate in the cloud even when devices are closed, allowing a single task to unfold over hours or even days.
Myriad: Predict which company will lead in AI models!.A standout feature of this new offering is the ability to create coworker agents. Managers can define roles, such as an events coordinator, prompting the Gemini to generate an agent complete with its own Workspace account, including an email address, calendar, Drive storage, and a listing in the company directory.
This concept mirrors initiatives from other platforms; for instance, Grok popularized this with Grok Bot, while OpenAI introduced similar functionalities with GPT dots, and Hermes launched its bots feature as well.
The Gemini Agent is capable of connecting to various platforms like Salesforce, ServiceNow, Jira, Snowflake, and BigQuery, utilizing the Model Context Protocol, which serves as a universal connection between AI and software. It also operates across multiple AI models, including Gemini and Claude from Anthropic, optimizing each task for quality and cost.
While the autonomy of these agents raises safety concerns, Google addresses this by treating each agent as an individual employee. Each agent is assigned a cryptographically verified identity, ensuring that their actions are logged in an audit trail attributed to the agent rather than a human operator.
As with any AI, cost management is crucial, as agents utilize tokens—units of text processed by AI models for which customers are billed. Google noted that nearly 500 clients processed over one trillion tokens collectively last year and has implemented real-time spending limits; when a project reaches its cap, the agent pauses until it is manually resumed.
Google has yet to announce the pricing or general availability for the Gemini agent. Currently, financial services and legal versions are available for preview, with government, healthcare, and retail versions anticipated in the near future.