Google launches Gemini agent for work that also runs on Anthropic's Claude models

Google Cloud announced a "Gemini agent" for businesses on Thursday, October 8, 2026, at its Gemini at Work event, describing it as a universal agent for work. The agent can plan tasks and carry them out across business applications, and it chooses the most suitable AI model for each task. Unusually for Google, it runs not only on Google's own Gemini models but also on Claude models from rival AI company Anthropic, with support for other private and open models planned. The agent is in private preview, and Google says it will be widely available soon to Workspace customers on select Business and Enterprise plans. Google plans to bring it to businesses first and to consumers later.

The big AI companies are racing to move from chatbots that answer questions to agents that do work: drafting documents, pulling data from several systems, filing requests and following up. Businesses are where that shift is happening first, because they have repetitive processes and the budgets to pay for automation. Google's announcement follows OpenAI's launch of always-on agents and Meta's expansion of its business agents in recent weeks, which shows how crowded this market has become.

The most striking choice is Google's decision to include Anthropic's Claude models inside its own product. Big platforms usually push their own technology. By letting the agent pick between Gemini and Claude, Google is telling corporate customers that the agent and its connections to their apps matter more than which model sits underneath. It also reflects what many companies want: the freedom to use more than one model provider rather than being locked into one.

Control remains with administrators. Keeping third-party models off by default and requiring an explicit opt-in addresses concerns many firms have about where their data goes and which vendors process it.

For India, where large IT services firms build and run AI systems for global clients, model-neutral agents could change how projects are designed. Teams may spend less effort tying their work to a single model and more on connecting agents to clients' business systems, data and security rules.

Short-term: Workspace business customers will be able to try the agent as it moves from preview to wider release, and rivals will face pressure to offer similar multi-model choice.

Long-term: If customers embrace agents that switch between models, AI model makers may compete more on price and specific strengths, while platforms compete on the agent layer and app integrations.

Who is affected: Businesses using Google Workspace and Google Cloud, IT teams and administrators, AI model providers such as Anthropic and Google DeepMind, and IT services firms that build enterprise AI.

Questions and Answers