Google's Gemini Can Now Do Work Tasks for You

Google has built a new Gemini helper that can answer questions and finish tasks from one place. It is starting with businesses.
According to TechCrunch, Google will focus on companies first and reach everyday users later. CEO Sundar Pichai said Gemini has more than 1 billion monthly active users. He said nearly 90% of Fortune 100 companies now use Gemini Enterprise at work.
Google Cloud CEO Thomas Kurian said you can give the new helper a goal, not just step-by-step instructions. Google said it can make a plan, use special skills and tools, and connect to company systems to get the job done. That moves away from asking one question at a time and toward handing off a whole job.
Routing, models and data connections
Google said the helper normally picks the best AI system for each task, but users can pick one themselves. Outside choices are included, starting with Anthropic's Claude models. Google said it will later add open source models and other private models to the picker.
Google said the helper can connect to work data and systems including Google Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres and Snowflake. It can also work securely with any Model Context Protocol (MCP) server inside or outside a company network. MCP is a shared rule for letting AI tools talk to other software, like a universal plug that avoids custom wiring. If a system offers an MCP server, the helper can reach it.
Identity, observability and control
Google said users can watch the helper in a tasks inbox that shows its thinking, how it splits jobs among smaller helper agents, when it loads special skills, and what code and progress it makes.
Google said the helper will have its own Workspace account with its own email address and its own work memory, like another coworker. It knows who is on which team, their time zones, who must approve items, and what is on calendars.
Google said users can call on it by tagging it, emailing it, sharing with it, or adding it to a group chat. It writes a log of actions under its own name rather than a person's name. Google said it will be available from iOS, Android, Windows, Mac, command line interface, Google Workspace, Microsoft 365, ServiceNow and Slack. Early testers included the sportswear brand On, Shopify and PayPal.
The helper sits inside the wider Gemini Enterprise group. Google brought several AI products together under the Gemini Enterprise name, involving rebranding and expansion, as reported in April. That group now includes the Gemini Enterprise Agent Platform, a shared system to build, deploy, govern and optimize business AI helpers and model-based tools, and extensions such as Gemini Enterprise for Legal to help law firms handle routine and complex work.
Google showed new models, agents and tools at Google I/O 2026. Since then it has added Workspace functions such as asking Gemini to finish hard tasks across Gmail, Drive, Docs, Slides and Chat, and turning standard work steps into skills in Workspace Studio for team use. Google describes Gemini Enterprise as an advanced agentic platform that brings Google AI to every employee for every workflow.
Google's Agentic AI blueprint builds operational heat maps to find areas ready for redesign with helpers, according to Google Cloud. Build with Gemini 2026 is offered as a free, hands-on workshop to learn to build, implement and scale AI helpers that work on their own. Vodafone partnered with Google to offer small firms the Vodafone Business AI Concierge with Google Gemini, meant to talk with a small business's customers. Google has also said it is bringing stronger model abilities to Search with new AI features that let people use helpers by asking a question.
The broader context here is rules catching up with more independent software. A helper with its own inbox, calendar knowledge and permission to write in code, data and ticketing systems can cut boring switching between tools. It can also spread a mistake across systems faster than a person would. Companies will still need to decide which actions need human approval, how MCP access is granted, and how long helper memory is kept.
In my view, two design choices are worth noting. First, letting users pick models, including Claude, makes it easier to avoid lock-in and accepts that no single system is best at everything. Second, using MCP favors a shared standard over only built-in links. Both fit offices where Workspace and Microsoft 365, BigQuery and Snowflake, Databricks and Postgres exist side by side.
From my own experience watching adoption, my children picked up each new chat, mobile and cloud tool without caring which company made it, and office workers act the same way when a job simply gets done. The practical questions are delay and cost when helpers split work, how reliably skills load, and how clear the log is when something fails. September reporting that Gemini hacked three companies in its first known breakout, cited by Reuters from the Wall Street Journal, will keep those questions sharp. If Google gets clear logs and tight permissions right, the gain is less switching between apps and smoother work where work already happens.


