Oracle Google Gemini deal brings models to Fusion AI agents

Oracle plans to add Google’s Gemini models to Oracle AI Agent Studio, giving Fusion Applications customers another model option for building AI agents. The Oracle Google Gemini expansion matters because Fusion is the strategic path for many existing Oracle application customers, and model availability inside that environment can shape which AI tools enterprises actually adopt.

The move extends Oracle’s partnership with Google Cloud. Oracle said Gemini will be available in AI Agent Studio, where customers and partners can create agents and agent-based applications tied to Fusion. Oracle also plans to use Gemini for embedded AI features in Oracle Fusion Applications and Oracle NetSuite.

Oracle identified Gemini 3.1 Flash Lite and Gemini 3.5 Flash as the models coming to AI Agent Studio. The company said the integration will add broader multimodal capabilities, a term that generally refers to systems that can work across formats such as text, images, audio or other inputs. Oracle did not provide rollout timing, pricing, adoption targets or financial terms for the expanded partnership.

What is Oracle adding Gemini to?

Oracle is adding Gemini to AI Agent Studio for Fusion Applications, its environment for building AI agents that work inside Oracle’s business application suite. Those agents are meant to automate tasks or assist with decisions in enterprise workflows, although the practical value depends on how tightly the agents are controlled, audited and connected to business data.

Google Cloud vice president Satish Thomas said in a prepared statement that the partnership is meant to make Gemini easier to use in the applications and agent workflows companies already depend on. Oracle applications development executive vice president Chris Leone said the addition gives customers and partners more choice when they build and extend agents for business problems.

The model-choice angle is the operative point. When Oracle introduced its LLM-powered agent platform for Fusion, it said customers could use models from Cohere and Meta and connect other supported models through the platform. Oracle has now put Google’s flagship AI model family on that menu. Whether the earlier Cohere and Meta options remain unchanged was not answered in the details announced.

The expansion also puts Oracle’s application AI strategy more directly in the same channel as other enterprise software vendors pushing agents into core systems of record. For customers, the question is less whether a vendor can expose a large language model through an application and more whether the resulting agent can be governed well enough for finance, HR, supply chain or customer operations.

Gartner has already urged caution on Oracle’s agent rollout. After Oracle launched Fusion Agentic Applications earlier this year, Balaji Abbabatulla, a Gartner vice president and vendor lead analyst for Oracle, said the pitch sounded promising but that implementation issues remained unresolved. He also raised the liability problem: if an AI agent makes a poor decision quickly and at scale, mistakes could spread before a human catches them.

Oracle’s answer so far has centered on monitoring and audit tooling, according to Gartner’s account. Abbabatulla said he still had not seen a clear answer from vendors on who is responsible when AI-driven decisions go wrong. That remains the harder enterprise sales problem behind the Gemini integration, especially for buyers being asked to put agents closer to operational decision-making.

This story draws on original reporting from The Register.