Fortanix recently launched a public preview of Armet AI, a turnkey generative AI platform that integrates confidential computing and enterprise-grade data governance.

Targeted at enterprises managing sensitive or regulated data, Armet AI addresses the challenges of building secure and compliant generative AI systems by combining Intel’s trusted execution environments (SGX, TDX) with fine-grained access control, policy enforcement, and AI-specific security mechanisms.

Key Capabilities

Armet AI is a secure and controlled platform for building custom generative AI applications that utilize sensitive or internal data. At its foundation, the platform leverages confidential computing, using Intel’s suite of technologies, to establish trusted execution environments that protect data during processing.

Confidential Computing Infrastructure

Fortanix integrates multiple layers of Intel’s hardware-based security technologies to deliver robust protection for data-in-use:

• Intel SGX (Software Guard Extensions) enables fine-grained memory encryption at the application level. This prevents unauthorized memory reads from even privileged system components.
• Intel TDX (Trust Domain Extensions) provides VM-level isolation and encryption, particularly useful for scaling confidential workloads in virtualized or containerized environments.
• Intel Tiber Trust Authority adds centralized attestation and policy validation, ensuring workloads are only executed in verified, trusted environments.

Data Governance and Access Control

To manage data access and maintain tight control over sensitive workflows, Armet AI provides an enterprise-grade policy engine:

• Organizations can define role-based access control (RBAC) policies governing who can view, query, or fine-tune data or models.
• Access is enforced at the application level and tightly integrated with the execution environment, ensuring consistency between logical access controls and physical data security.

AI Guardrails and Content Filtering

Security extends into the AI interaction layer, where Fortanix introduces proactive measures to prevent unintended data exposure and harmful outputs:

• Automatic redaction of PII and other sensitive attributes from both user prompts and AI-generated outputs.
• Input and output filtering that screens for policy violations, adversarial prompts, and content that could introduce reputational or legal risk.

Integration, Observability, and User Experience

The platform includes tools to simplify deployment and provide governance transparency:

• REST APIs and prebuilt connectors facilitate data ingestion and integration into existing enterprise systems, reducing development overhead.
• A secure chat interface allows employees to interact with LLMs based on internal data while maintaining strict data control.
• A real-time security dashboard provides monitoring, alerting, and audit trail capabilities to support internal governance and external regulatory requirements.

Impact to IT Teams

The introduction of Armet AI can reduce implementation complexity and operational risk for IT operations, security teams, and developers. Enterprises that have hesitated to deploy AI due to data governance or compliance constraints now have access to an integrated solution that abstracts the security architecture while enforcing strong controls.

From an operational perspective, this offers the following benefits:

• Reduced internal complexity: Enterprises can bypass building custom secure inference pipelines or managing encryption logic by offloading those responsibilities to Fortanix’s confidential computing stack.
• Improved audit readiness: Integration of real-time monitoring and role-based control simplifies compliance with data privacy regulations such as GDPR, HIPAA, and PCI-DSS.
• Controlled LLM usage: Unlike public SaaS-based AI offerings, internal data does not need to leave the enterprise perimeter, and access can be tightly scoped per role or project.

Analysis

Fortanix is carving out a focused position in the secure generative AI segment in a market increasingly populated by general-purpose AI tools. Armet AI differentiates itself through its security-first design and deep integration with confidential computing infrastructure. These capabilities are ideal for enterprises that require data privacy and sovereignty guarantees.

Fortanix’s strategy is to fill the void between open-source LLM deployments, which can be difficult to secure, and SaaS AI offerings, which may not meet compliance standards. Providing a turnkey solution that balances usability with enterprise-grade control allows Fortanix to enable compliant AI deployment in sensitive industries.

Armet AI arrives when enterprises are actively evaluating how to reconcile the promise of generative AI with the constraints of data protection, regulatory compliance, and internal governance.

By grounding its platform in confidential computing and extending its feature set with enterprise-grade policy controls and observability, Fortanix offers a compelling value proposition for regulated industries and data-sensitive applications.

Competitive Outlook & Advice to IT Buyers

These sections are only available to NAND Research clients and IT Advisory Members. Please reach out to info@nand-research.com to learn more.