Gemini Enterprise is Google’s new ‘front door’ for agentic AI access at work

Google has introduced a new AI assistant to serve as a platform so users can access and coordinate agents that automate work tasks. Gemini Enterprise, which replaces the Agentspace app launched last year, also features new enterprise search functions to help customers tap into data from across an organization’s business apps. 

Gemini Enterprise is aimed at a wider range of business users and serves as the “new front door for every user to use AI in the workplace,” Google Cloud CEO Thomas Kurian said in a briefing. It enables users to “chat with all their enterprise data, search for information and use agents to do a variety of tasks on their behalf.” 

The app features a chat interface similar to Google’s existing Gemini AI assistant. It’s built on Google’s Gemini large language models (LLMs) and provides access to the company’s latest generative AI (genAI) media models such as Imagen, Veo, and Gemini 2.5 Flash (also known as “nano banana”). 

With Gemini Enterprise, business users can find, build and interact with available agents via the chat interface.  

There are several categories of agents, including Google’s own “core” agents, such as NotebookLM and Deep Research, as well as the new Code Assist agent, which launches today, and the upcoming Data Science and Customer Engagement Suite agents. The likes of Box, Salesforce and ServiceNow have built agents that are available from Gemini Enterprise, and users can access custom agents built from scratch in Google Vertex. 

Google’s visual no-code Agent Designer tool (currently available in preview) lets users build and configure their own agents in Gemini Enterprise. 

For example, a “campaigns agent” can be set up to coordinate the processes for a marketing campaign using multiple agents to automate tasks such as identifying trends and performing competitor analysis; generating marketing copy, video and images; drafting communications; and checking inventory.

A central governance framework lets admins manage and monitor agents that have been deployed in their organization. Google’s Model Armor is also included in Gemini Enterprise to screen user prompts and help protect against risks such as prompt injections and data leakage.

Gemini Enterprise also serves as an enterprise search tool that lets users locate information held across anorganization’s enterprise systems, whether that’s Google’s own product or apps from third-parties (subject to permissions being granted). It ensures that responses from Gemini Enterprise are grounded in data from ERP and CRM apps, enterprise documents, databases, and so on. 

To simplify integration, Google provides several “packaged connectors” to hook Gemini Enterprise into these systems, Kurian said, including Box, Salesforce, and Microsoft Teams. (Google posted more information online about the offering.)

“By ensuring agents are grounded in unique organizational data and connected to critical business applications, Google Cloud is tackling the architectural hurdle that a majority of organizations struggle with,” said Mike Leone, practice director for data, analytics and AI at Omdia’s Enterprise Strategy Group. “They are solving integration headaches before they even begin.”

The launch of Gemini Enterprise marks the reintroduction of a premium AI add-on for Google Workspace customers. Last year, Google opted to include its Gemini AI assistant and AI features such as summarization and text generation in apps such as Gmail and Docs to Workspace customers at no extra cost. (The overall price of the subscription, however, was simultaneously increased.)

Google offers several SKUs: Gemini Enterprise Standard and Plus editions cost $30 per seat per month, while Business edition will cost $21 per seat a month. The Business version will begin rolling out to customers Oct. 10, Google said; it provided no information on the Standard and Plus editions. 

The Gemini Enterprise announcement comes as many business software vendors are integrating AI agent builder and orchestration tools into their products. This includes Microsoft 365 Copilot and Copilot Studio, Atlassian’s Rovo, and Asana’s AI Teammate, to name a few. OpenAI recently announced a range of tools for building AI agents based on its ChatGPT models. And there are specialized agent orchestration tools, including n8n, Make and Airflow.

“The market as a whole has entered a high-stakes race, where every major player is chasing agentic AI orchestration, recognizing that autonomous and intelligent systems are the next great leap in productivity,” said Leone.

Google’s strength lies in its ability to help businessses scale agentic systems securely — a major stumbling block for adoption. “Its core differentiation is the strategic decision to be a centralized ‘single pane of glass’ for security, auditability, and visualization of all AI agents,” said Leone. 

One challenge for Google Cloud is helping customers avoid confusion when accessing its broader AI portfolio. “The sheer array of product names, services, and tiers within their AI ecosystem can introduce unnecessary friction,” said Leone. 

“If Gemini Enterprise is being positioned as the definitive entry point to a customer’s agentic AI future, Google Cloud needs to execute a strategy that eliminates potential customer confusion that will maximize adoption by making the path to scaled, trusted AI as clear as possible,” he said.Gemini Enterprise is Google’s new ‘front door’ for agentic AI access at work – ComputerworldRead More