Google launches Agentspace. It is a solution that securely combines Gemini with company data, enabling research and action taking.
Google presents Agentspace. Agentspace is a customized solution for businesses, where Google combines its Gemini LLM with its own search technology, company data and the ability to take action.
In fact, it is a slightly expanded version of Gemini, but the evolution of AI is not only driven by actual progress, but also by buzzwords. This time, every update to an AI capability apparently has to have the word “agent” in its name, and Google doesn’t want to be left behind.
Reasoning from prompts and business data
Agentspace still works based on natural language prompts, but is able to reason further about these prompts. The AI solution can search for relevant information in company data, wherever it is located. This allows employees to surface all relevant information with a single prompt, without having to manually search through different systems. Agentspace works with structured and unstructured data, even when documents are available in different languages.
Google says Agentspace can link to Confluence, Jora, Microsoft SharePoint, ServiceNow and its own Google Drive, among others, at launch. Agentspace and Gemini 2.0 are also included in the new NotebookLM Plus, which is available in early access.
Agents
Google Agentspace also integrates with systems and APIs to take action upon a user’s request. Low-code solutions will allow employees to automate workflows and then deploy those automations through the Agentspace interface. Google shows a demo where Agentspace summarizes tickets from Jira and sends an email via Outlook based on those tickets, as well as another example where the tool helps create an output report.
Google argues that Agentspace is relevant to all profiles within a company. The solution can, according to the company:
- Software engineers contribute to this (but also Gemini and other artificial intelligence tools);
- Support marketers (as before, but with faster access to company data);
- Analysts help spot trends (again by linking company data);
- Assist human resources specialists (with automation, for onboarding, for example).
Google insists that Agentspace is secure, with verifiable access to company data and compliance guarantees. Agentspace is available in early access.
Agentspace, Agentforce et Copilot
The above doesn’t seem all that new. First of all, the name (which is probably not a coincidence) is strongly reminiscent of Salesforce’s Agentforce. Agentforce, announced at Dreamforce, fulfills a similar role, but within the Salesforce ecosystem. It’s also worth noting that the functionality Google is unpacking with is reminiscent of Microsoft’s Copilot integrations. Copilot, combined with Power Apps, for example, already brings AI and automation together. Only Microsoft named this feature before the word “agent” became popular.
One thing is clear: while last year LLM-based AI was expected to provide passive assistance, the trend is now towards activation. AI must be present in multiple systems (or preferably all systems), in a secure manner, and be able to process complex prompts. Additionally, according to Google, Salesforce and Microsoft, users should be able to at least trigger simple tasks via the LL interface. Agentspace fits perfectly into this trend.
