The End of Shadow AI: AI/R Compass UOL Expands Enterprise Governance

SAN FRANCISCO – November 10, 2026 — In the rapidly evolving landscape of modern enterprise technology, a silent crisis has been brewing beneath the surface of corporate innovation. Across global organizations, well-meaning department heads and eager teams are quietly swiping corporate credit cards to access hundreds of disparate artificial intelligence models. This phenomenon, widely known as "shadow AI," operates outside the purview of central IT, creating a sprawling, unmonitored web of data privacy risks, compliance violations, and spiraling inference costs.

Today, AI/R Compass UOL—a technology subsidiary of AI/R Group US Holding Corp specializing in Agentic AI—announced a significant expansion of its AI/Cockpit One platform. The system now supports more than 80 AI providers and over 600 distinct artificial intelligence models. But beyond the sheer volume of integrations, the announcement signals a critical shift in how corporations must manage the systems that power their digital futures. By pairing broad model choice with an uncompromising security approval workflow, the platform acts as a digital border control, ensuring that innovation does not come at the expense of corporate responsibility.

Taming the Wild West of Enterprise AI

For the past three years, the dominant narrative in artificial intelligence has been one of unbridled access. The rush to integrate generative capabilities led to a fragmented ecosystem where sensitive corporate data was routinely fed into third-party models with opaque data retention policies. The risk of proprietary information inadvertently becoming training data for public models has kept Chief Information Security Officers awake at night.

AI/Cockpit One approaches this systemic vulnerability by operating on a principle of "default-off." Unlike standard aggregator tools that simply provide API keys to a buffet of models, the platform requires every model to pass a rigorous security vetting process before it is made available to enterprise users. This workflow specifically evaluates the provider's data retention policies and demands contractual or programmatic confirmation that corporate data will not be used for model training.

“It’s not enough to enable models and govern their use within the company, because exposure and leakage risks can also lie with the providers,” said Cleyton Ferreira, Director of Technology and Products at AI/R. “That’s why we designed a rigorous approval workflow: we only enable models that meet our security criteria, which include evaluating retention policies and requiring that corporate data not be used for training. Expanding support means more options, delivered responsibly, rather than indiscriminate access. Without a managed environment, each team picks its own provider, and the company loses visibility into how much it’s spending and where its data is.”

The necessity of this approach is echoed by cybersecurity professionals across the industry. One enterprise IT procurement leader noted that managing the lifecycle of AI data is no longer just a best practice, but a regulatory mandate. With frameworks like the EU AI Act specifying stringent retention periods for data in high-risk AI systems, platforms that can centrally audit and enforce these policies are transitioning from optional middleware to essential infrastructure.

The FinOps of AI: Smarter Routing for Sustainable Scale

Beyond the ethical and security imperatives, the expansion of AI/Cockpit One addresses a highly pragmatic concern: financial sustainability. The computational resources required to run advanced, frontier generative models are immense, and using these top-tier systems for mundane administrative tasks is akin to chartering a commercial jet for a grocery run.

By integrating over 600 models into a single managed environment, the platform enables intelligent, task-based model routing. Organizations can dynamically direct simple, routine tasks to faster, highly efficient, and less expensive models, while reserving the heavy computational power of advanced models for complex, high-stakes workflows. This flexibility not only accelerates innovation but directly mitigates the risk of vendor lock-in, allowing companies to pivot away from providers that raise prices or alter their privacy terms.

Furthermore, the platform introduces rigorous financial governance to an area previously characterized by unpredictable billing. Through centralized administrative features, AI/Cockpit One tracks token consumption, generates model- and user-level cost reports, and enforces team-based spending limits. For FinOps practitioners trying to balance the books, this granular visibility transforms AI from a black-box expense into a measurable, optimizable investment.

The Systems Integrator Pivot and the AWS Ecosystem

The launch also highlights a broader shift in the technology services industry. Traditional IT consultancies and system integrators are increasingly pivoting away from purely service-hour billing models, choosing instead to build and license proprietary software layers. By developing robust middleware like AI/Cockpit One, service firms are capturing more enduring value and differentiating themselves in a crowded cloud market.

This strategy is particularly evident in the firm's deep alignment with Amazon Web Services. As an AWS Premier Partner holding 12 technical competencies and employing over 1,900 AWS-certified professionals, the company has won the AWS SI Partner of the Year award for Brazil and Latin America five times. However, rather than competing with AWS's native tools, the platform is designed to complement them.

While AWS Bedrock provides a curated, highly secure environment for foundational models, AI/Cockpit One acts as an overarching orchestration layer. It bridges the gap between AWS-native models and the broader ecosystem of 80+ external providers, offering centralized role-based access control (RBAC), Single Sign-On (SSO), and unified audit logging across chats, agents, and applications. This positions the technology not as a replacement, but as an essential companion for enterprises navigating hybrid, multi-cloud, and multi-model realities.

Building Responsible Digital Infrastructure

As we look toward the collective future of our digital society, the systems we build today will dictate the safety and equity of tomorrow's technological landscape. The unchecked proliferation of "shadow AI" represents a systemic failure of governance—a scenario where speed is prioritized over the ethical stewardship of data.

The expansion of platforms that mandate security approvals, enforce data privacy, and provide transparent financial telemetry is a necessary maturation of the AI industry. It reflects a growing recognition that true corporate responsibility requires more than just publishing ethical guidelines; it requires hardcoded digital boundaries. By demanding that every artificial intelligence model prove its safety before it is allowed to operate within the corporate walls, we are taking a vital step toward a technological ecosystem that serves human needs without compromising our fundamental rights to privacy and security.