Google launches Gemini subscriptions to help corporate workers build AI agents

Google is taking another shot at selling businesses on artificial intelligence agents by introducing subscriptions featuring agents that perform specific work tasks.

Gemini Enterprise targets large organizations, starting at a monthly fee of $30 per person. Gemini Business, for smaller clients, costs $21 per person each month. The offerings enable corporate workers to build agents that draw on data from Box, Microsoft and Salesforce products.

Premade Google agents for software development, data science and customer engagement also come with the new Gemini subscriptions, along with access to agents from Workday and other companies. They include the capabilities of Agentspace, an agent building product Google announced in December. Google will upgrade current Agentspace clients to Gemini Enterprise or Gemini Business free of charge through the course of their contracts, a spokesperson said.

Gemini subscriptions come with Model Armor, a feature for inspecting and blocking requests and responses in AI chats, so organizations don’t need to fuss with setting it up.

The launch comes three days after OpenAI showed how people can access tools from third-party apps in ChatGPT. Google and Microsoft, meanwhile, are looking to get enterprises hooked on agents that take care of some processes, so employees can do other things. Both companies sell services aimed at developers and at nontechnical workers. Neither Gemini Enterprise nor Gemini Business require coding.

“We’ve seen people from consulting services companies, telecommunications companies, software companies, hospitality companies and a variety of different manufacturing companies all using these, and in a variety of scenarios,” said Thomas Kurian, CEO of Google’s cloud group, in a media briefing.

Kurian, who accelerated the unit’s year-over-year revenue growth back above 30% in the second quarter, named cruise line Virgin Voyages as a Gemini Enterprise early adopter.

Firms are more likely to be exploring or testing AI agents than putting them into production, said Chirag Dekate, an analyst at technology industry researcher Gartner. But Google’s handling of security and governance should ease concerns among big companies evaluating agent systems, Dekate said.

Google’s new Gemini subscriptions depend on the company’s Gemini AI models for working with text, images and videos. Google and other model makers regularly release new versions, and enterprises want to avoid getting stuck with lagging models when selecting agent software, Dekate said.

“How Google is able to leverage this unified messaging in the Gemini 3.0 launch sequence, which is coming soon, I think, will also be a crucial litmus test,” he said. “In other words, will they be able to offer a same-day sort of innovation cycle, or is this going to be staggered in terms of adoption patterns?”

WATCH: Box CEO on AI monetization: Agents offer new monetization for existing software companies