Gemini Enterprise with Data in Brazil: What It Means for Your AI Strategy

For a long time, the question stalling generative AI projects at Brazilian companies wasn’t technical. It was legal: “Where will our data be processed?” For banks, insurers, healthcare companies, and regulated entities, that question alone was enough to push projects to the next quarter.

That barrier is gone. With Gemini Enterprise processing data in Brazil, the conversation moves up a level: from “Can we use AI?” to “How do we use AI strategically, with governance and measurable returns?” This article is for the people who need to answer that second question.

The announcement: data residency and infrastructure doubling by 2030

On September 24, 2026, at Google Cloud Summit Brasil, Google announced that the web version of Gemini Enterprise would process data in Brazil for customers using Gemini 3.5 Flash. The change took effect on October 15. Customer data for agent workflows was already stored locally; now processing stays in the country too.

In practice, this means three things for technology teams:

• Simpler compliance. Fewer international data transfers mean less friction with regulatory and sovereignty requirements, a sensitive issue for companies subject to Brazil’s LGPD data protection law and sector regulators.
• Low latency. Faster responses for use cases that depend on real-time interaction.
• Continuity. Local processing also strengthens high availability and disaster recovery strategies.

The announcement came with a long-term commitment: Google plans to double its technical infrastructure in Brazil by 2030 and has already made G4 virtual machines, with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, available in its São Paulo region for heavy inference and real-time workloads. Google also reaffirms that, in every region, customer data isn’t used to train its models without permission.

The message is clear: Brazil is no longer a market served “from abroad” but a priority region for agentic AI.

The real problem: fragmented AI, one copilot per app

Data residency solves the “where.” It doesn’t solve the problem most companies already face: AI arrived in a disorganized way.

The scenario is familiar to any CIO. The CRM has its own copilot. So does the HR tool. Marketing subscribes to a solution on its own, legal is testing another, and some employees use tools IT never approved. Each with its own contract, pricing model, and data policy.

The result is predictable:

• isolated productivity gains that don’t scale across operations;
• AI budget scattered across departments, with no consolidated view of cost or return;
• governance risk, with corporate data flowing through tools outside IT’s control.

The Work: In Progress study, conducted by Google Cloud with IDC and Provokers, shows the size of the gap: 59% of respondents in Brazil already delegate tasks to AI agents, but only 17% of companies have backed that adoption with the governance needed to scale agents across multiple core processes.

Thomas Kurian, CEO of Google Cloud, summed up the diagnosis at the event: customers report the same three challenges when scaling AI, namely cost, unclear business value, and the need for enterprise-grade controls.

This is where strategy changes. Gemini Enterprise shouldn’t be treated as “one more copilot” but as the company’s single AI platform: one place to search, analyze, create, and automate, connected to Google Workspace and the systems you already use.

Projects, Skill Registry, and Agent Sandbox: governance from day one

The new Gemini Enterprise capabilities target exactly the gap between adoption and governance:

• Projects. Dedicated spaces where an agent’s memory is limited to the files and conversations the team defines. Projects are persistent, keeping shared memory, context, and state over time. For IT, that means moving beyond one-off prompts and organizing AI use by initiative, department, or client, with controlled context.
• Skill Registry. A central, shared library of skills: packages of instructions, scripts, and resources that teach an agent to perform a specialized task. It’s the difference between dozens of improvised automations and a corporate catalog that the whole company can reuse.
• Agent Sandbox. A controlled environment for testing agents before they go into production, inside the company’s security perimeter and integrated with Agent Identity, Agent Gateway, and Agent Observability. For security teams, it’s what makes it possible to validate autonomous agents without exposing production systems.

Together, these capabilities let you design governance into the very first use case instead of patching it in after AI has already spread across the company.

Cases in Brazil: from pilot to results

Adoption is already underway at large Brazilian companies, with concrete numbers:

• Bradesco, one of Brazil’s largest banks, uses Gemini Enterprise for tax, accounting, and contract analysis. Document review time dropped from one hour to five minutes, and inconsistency-related risks fell 60%.
• Livelo, a Brazilian loyalty program, rolled out the platform in a self-service model, reaching about 92% internal engagement, and automated review moderation and its digital channel and campaign pipeline, recovering 20,000 hours a year.

Other examples cited by Google follow the same logic: Sabesp, São Paulo’s water and sanitation company, uses agents in its legal workflow to identify potential fines and speed up defense preparation, and insurer SulAmérica builds specialized, governed agents to free up hours spent on repetitive tasks such as tickets and reports.

The common pattern is clear: specific use cases with business metrics defined from the start. None of them began with “let’s roll out AI across the whole company.”

How to get started: assessment, prioritization, and incentive programs

If your company already uses AI agents but isn’t yet among the 17% that govern that adoption, the safest path is to start with a diagnosis, not a tool.

1. AI Productivity Assessment. A map of where productivity is being lost today (information search, reporting, content creation, document analysis) and which AI tools are already in use, approved or not. The result is an objective comparison between the cost of today’s fragmented AI and the return of a unified platform.
2. Use case prioritization. Based on the assessment, choose two or three use cases with clear metrics, as in the Bradesco and Livelo examples: review time, operating hours, cost per process.
3. Governed deployment. Set up Projects, Skill Registry, and access policies from the start, with Gemini Enterprise integrated with Google Workspace (or your current ecosystem) and your company’s data systems.
4. Adoption and ongoing operations. Enterprise AI doesn’t end at go-live. Adoption management, license optimization, and agent evolution are what turn a pilot into recurring results.

And here’s something many leaders don’t know: Google Cloud runs incentive programs, accessed through partners, that can co-fund AI value assessments, proofs of concept, and eligible deployment projects, such as the DAF (Deal Acceleration Fund). In practice, this lowers the upfront investment and the risk of taking the first step.

Questions to bring to your next board meeting

Before deciding, it’s worth answering these internally:

• Do we have an AI strategy that goes beyond chatbots?
• Where are we losing the most productivity: search, reporting, or content creation?
• Which AI tools are employees already using without IT approval?
• How much are we spending today across all the AI subscriptions scattered among departments?

If any answer is “we don’t know,” that’s your starting point.

Gemini Enterprise with Maplink Cloud

Maplink Cloud has spent more than 25 years bringing together geolocation, cloud, and data intelligence for companies across Latin America, with offices in São Paulo, Buenos Aires, Santiago, and Mexico City. As a Premier Partner for Google Cloud and Google Maps Platform, with a Cloud & AI First approach, we run the AI Productivity Assessment, help you access Google’s incentive programs, and deploy Gemini Enterprise with governance from day one, backed by local support in Portuguese and Spanish.

Want to understand what Gemini Enterprise with data in Brazil means for your company’s AI strategy?

Talk to a specialist ➜