Gemini Meets Oracle: Gemini Connector, Oracle AI Database Agent and Gemini in Fusion

On this page

1. 01 · Two Directions, One Partnership
2. 02 · What Was Announced, and Its Status
3. 03 · The Gemini Connector (Preview)
4. 04 · Oracle in Gemini: Available Today
5. 05 · Gemini in Oracle Applications
6. 06 · Partner Blueprints
7. 07 · Governance: Grounding in Knowledge Catalog
8. 08 · Which Option Fits Your Team?
9. 09 · Key Takeaways
10. 10 · Frequently Asked Questions

01 · Two Directions, One Partnership

Oracle AI World, Las Vegas · 28 October 2026

Ask Gemini about Oracle data, or bring Gemini into Oracle apps.

For years, getting an answer out of an Oracle system meant a report request, a SQL query or a dashboard someone had already built. Google Cloud's 28 October 2026 announcement for Oracle AI Database@Google Cloud describes a different path: ask in plain language and let an AI agent find, query and explain the data, while the governance rules you already set decide what each person may see.

The announcement runs in two directions. Oracle in Gemini brings Oracle data and Oracle-aware agents into Gemini, where business users and developers already work. Gemini in Oracle applications embeds Gemini's frontier reasoning inside Oracle Fusion AI Agent Studio, Fusion Claw, Oracle Fusion Applications and NetSuite. A third piece, the Gemini Connector for Oracle AI Database@Google Cloud, lets the Gemini app query Oracle databases directly, grounded in Knowledge Catalog. That one is in preview.

02 · What Was Announced, and Its Status

Answer: Oracle in Gemini is available today. The Gemini Connector for Oracle AI Database@Google Cloud and the Knowledge Catalog integration are in preview and need sign-up. Gemini in Oracle applications is described as embedding Gemini reasoning in Fusion and NetSuite; the post gives no separate GA date for it.

Announcements like this mix shipped features with previews, and it is easy to lose track of which is which. The table below sorts them. Plan production work only on the rows marked available or GA, and treat previews as pilots that can change before general availability.

03 · The Gemini Connector (Preview)

Answer: The Gemini Connector for Oracle AI Database@Google Cloud lets people ask questions about Oracle data in plain language inside the Gemini app. Gemini uses Knowledge Catalog to discover, search and ground its answers strictly in the governed Oracle data each user is authorized to access. It is in preview, with sign-up.

Picture a finance manager who wants to know which suppliers had the most late invoices last quarter. Today that is a ticket to the reporting team. With the Gemini Connector, the manager types the question into the Gemini app. Gemini looks up in Knowledge Catalog which Oracle tables and columns describe suppliers and invoices, forms the query, and answers in plain language.

The important words in Google's description are "strictly in the governed Oracle data each user is authorized to access". The connector is not a back door around your access model. Two people asking the same question can get different answers, because each sees only what they are allowed to see. That is what makes it usable for real business data rather than demo schemas.

What we do not know yet

The announcement does not describe which Gemini editions include the connector, how it is licensed, which Oracle database services it supports, or how authorization is mapped from a Gemini user to Oracle data. Those details matter for any rollout. Sign up for the preview and check Google's documentation rather than assuming.

04 · Oracle in Gemini: Available Today

Answer: Oracle in Gemini brings conversational analytics, cross-cloud database connectors and domain-aware agents, including the new Oracle AI Database Agent, into Gemini. It ships with over 135 prebuilt operational and developer skills, uses standardized MCP connectivity with zero plaintext credentials, and supports multi-agent workflows across JD Edwards, E-Business Suite, NetSuite and Oracle Fusion.

This is the part of the announcement you can use now. Instead of a single chatbot, Oracle in Gemini is a set of capabilities. Conversational analytics answers business questions. Cross-cloud database connectors reach Oracle data. Domain-aware agents, led by the Oracle AI Database Agent, know enough about Oracle to act on it. And a library of more than 135 prebuilt skills covers common operational and developer tasks, so teams start from working building blocks instead of blank prompts.

The figure below places these pieces in one picture. Users enter on the left: business users in the Gemini app, developers building agents, and Oracle application users. In the Google Cloud area, Gemini and the Oracle AI Database Agent reach Oracle data through MCP connectivity, with Knowledge Catalog providing grounding. On the right, the Oracle applications take part in multi-agent workflows, and Gemini reasoning flows into Fusion AI Agent Studio and Fusion Claw.

Two flows stand out. The black arrows are Oracle in Gemini reaching data through MCP, and the orange arrows fan out to the four Oracle applications for multi-agent workflows. The purple dashed arrow runs the other way: Oracle applications calling Gemini for reasoning. Knowledge Catalog, on the red dashed link, keeps answers grounded.

The MCP piece links back to the previous post. The same principle applies: agents connect through a standard protocol, and credentials never appear in plain text. Google does not list the 135 skills or describe each agent's permissions, so review the published skill catalog before you enable anything in production.

05 · Gemini in Oracle Applications

Answer: Gemini's frontier reasoning is being embedded in Oracle Fusion AI Agent Studio, Fusion Claw, Oracle Fusion Applications and NetSuite, so agents inside those applications can automate execution across systems of record. Here Oracle's apps call Gemini, the reverse of Oracle in Gemini.

Many employees never open the Gemini app. They live in Oracle Fusion for finance, procurement, HR or supply chain, or in NetSuite for running a growing business. For them, the useful change is Gemini inside the tools they already use. Oracle Fusion AI Agent Studio is where agents for Fusion are designed and run, and the announcement adds Gemini's reasoning to it, to Fusion Claw, to Fusion Applications and to NetSuite. The goal, in Google's words, is to automate execution across systems of record.

The figure below puts the two directions side by side as layered stacks. Each layer answers one question: where users work, who reasons and acts, how they connect, and what holds the data. The left stack is Oracle in Gemini, where Gemini calls into Oracle. The right stack is Gemini in Oracle, where the Oracle application calls Gemini for reasoning and then acts on its own data.

Read across each row to compare. The user surface differs (Gemini app versus Oracle app), the reasoning engine is Gemini in both cases, and the data is Oracle in both cases. What changes is who owns the conversation and the workflow, and therefore which team configures and governs it.

Questions to ask Oracle

The announcement names the products but not the packaging. Before planning, ask your Oracle account team which Fusion and NetSuite releases include Gemini reasoning, whether it is generally available in your region, how it is licensed, and how data handling is governed when Fusion calls Gemini.

06 · Partner Blueprints

Answer: Ten partners, Argano, Automus, Calfus, HCL, IBM, Infosys, LTM, Pythian, TCS and Tessell, provide prebuilt skills, custom MCP connectors and industry blueprints for Gemini and Oracle deployments.

Most enterprises will not build every agent themselves. The announcement names a partner ecosystem that packages the common work. Their offerings fall into three kinds, and it helps to know which one you are buying.

The named partners are Argano, Automus, Calfus, HCL, IBM, Infosys, LTM, Pythian, TCS and Tessell. The announcement does not describe each partner's specific offerings, so ask for concrete skill lists, the data each connector touches, and how they handle credentials and permissions.

07 · Governance: Grounding in Knowledge Catalog

Answer: Knowledge Catalog, in preview, catalogs Oracle schemas and business metadata and applies consistent access controls, data provenance tracking and semantic context. The Gemini Connector uses it to ground answers strictly in the Oracle data each user is authorized to access.

A language model that answers confidently from the wrong table is worse than no answer. Grounding means tying each answer to specific, governed data the user is allowed to see. Knowledge Catalog plays that role. It describes what each Oracle table means in business terms, tracks where data came from, and keeps access rules consistent across Oracle databases and Google's Agentic Data Cloud.

For governance teams, this is the piece to study first. If the catalog's business descriptions are poor, agents will pick the wrong data. If access controls in the catalog do not match your database policies, users will see confusing gaps. Good metadata and aligned permissions are the groundwork that makes every Gemini and Oracle feature trustworthy.

Gemini can only be as right as the data it is allowed to see and the descriptions it is given. Governance is not a brake on AI here; it is what makes the answers worth trusting.

08 · Which Option Fits Your Team?

Answer: Business users asking questions start with Oracle in Gemini today and pilot the Gemini Connector if preview is acceptable. Developers build on the GA Remote MCP Server, skills and partner blueprints. Fusion and NetSuite users get Gemini through Oracle's applications.

The right starting point depends on who will use it and where they already work. The decision flow below starts from that question, then separates generally available features from previews for the business-user path, because that is where the two are easiest to confuse.

Most organizations will end up on more than one path: analysts in Gemini, developers building custom agents and finance teams in Fusion. Start with one path that has a clear owner and a measurable question to answer, then extend.

1. Pick one audience and one question. For example, finance managers asking about late supplier invoices.
2. Check status. Use what is available today for production; sign up for previews only as pilots.
3. Prepare metadata. Make sure the Oracle tables involved have clear business descriptions and owners.
4. Set least-privilege access. Confirm each user or agent sees only the data it should.
5. Run a pilot with real users. Compare answers with existing reports and record mismatches.
6. Decide and extend. Add audiences, skills or partner blueprints once the first path is trusted.

09 · Key Takeaways

1. Oracle in Gemini is available today. Conversational analytics, cross-cloud connectors and the new Oracle AI Database Agent with 135+ skills.
2. The Gemini Connector is in preview. It answers plain-language questions in the Gemini app, grounded in governed Oracle data via Knowledge Catalog.
3. Gemini also goes into Oracle apps. Fusion AI Agent Studio, Fusion Claw, Fusion Applications and NetSuite gain Gemini reasoning.
4. Partners package the work. Ten named partners offer prebuilt skills, custom MCP connectors and industry blueprints.
5. Governance comes first. Good metadata and least-privilege access decide whether answers can be trusted.

10 · Frequently Asked Questions