Virtana announced support for AWS Bedrock Guardrails within Virtana AI Factory Observability (AIFO), extending behavioral observability across enterprise LLM deployments on AWS Bedrock. 

As organizations adopt generative AI for mission-critical workflows, the operational challenge shifts from deploying models to operating them securely at scale. AWS Bedrock Guardrails provides the enforcement layer, blocking harmful content, masking PII, and defending against prompt injection. Virtana AIFO delivers the intelligence layer, making Guardrails activity observable and surfacing the behavioral patterns that distinguish legitimate workloads from active adversarial campaigns. Together, they give enterprises the defense in depth required to operate AI with confidence in production, part of Virtana's continued expansion of AI Factory Observability across the environments where enterprises deploy, run, and secure AI at scale.

“Enterprises are making significant investments in generative AI across an expanding range of environments, and the governance expectations around those investments are rising fast,” said Paul Appleby, CEO of Virtana. “Running AI in production means being accountable for how it behaves wherever it is deployed. Virtana AIFO gives security and operations teams the operational intelligence to meet that standard across infrastructure, platforms and LLMs and services like AWS Bedrock.”

AWS Bedrock Guardrails addresses content-level risk with a comprehensive, configurable set of safeguards that integrate directly into the generative AI workflow, filtering harmful content, masking PII, enforcing denied topics, validating contextual grounding, and running automated reasoning checks. These controls operate consistently across model inference, agents, knowledge bases, and multi-step flows, giving organizations a model-agnostic enforcement layer across their Bedrock environment. As enterprises run multiple foundation models through Bedrock for distinct workflows, maintaining consistent governance across each becomes an operational requirement in its own right.

Virtana AIFO monitors LLM behavioral patterns across AWS Bedrock deployments, treating every token pattern, utilization shift, and request anomaly as a potential security signal. When Guardrails intervention rates spike, when prompt token volume surges outside normal operating bounds, when request failure rates climb against a specific model, those patterns carry intelligence. They signal whether an anomaly reflects a configuration issue, a performance degradation, or an organized adversarial campaign testing the boundaries of enterprise AI defenses. Virtana AIFO surfaces that signal in real time, connecting Guardrails activity to the full behavioral context of the LLM environment.

Virtana addresses this operational requirement by delivering:

• Guardrails intervention monitoring tracks trigger frequency, blocked-topic patterns, and intervention rate trends by model, so security teams can detect active adversarial campaigns, not just individual blocked events
• Token-level behavioral analysis monitors prompt and completion token volumes, Time to First Token (TTFT), and request throughput to surface anomalous consumption patterns that indicate adversarial probing or data exfiltration attempts
• Request failure rate tracking identifies elevated failure rates as signals of credential misuse, adversarial probing, or Guardrails evasion activity across foundation model deployments
• Historical trend analysis correlates token volume spikes and Guardrails trigger patterns with known events or surfacing unknown anomalies, giving operations teams the context to distinguish legitimate workloads from active threats
• Cross-model visibility provides a unified operational view across all foundation models in the Bedrock environment to detect behavioral anomalies and support consistent AI governance across the enterprise LLM estate
• On-premises deployment with tenant-level data segregation and support for customer-managed LLM models meets the data sovereignty and compliance requirements of regulated industries including healthcare, financial services, and government

“Agentic AI systems introduce attack surfaces that content-level enforcement alone cannot address,” said Amitkumar Rathi, Chief Product Officer at Virtana. “By extending AI Factory Observability into AWS Bedrock environments, we give organizations visibility into the behavioral layer that sits above content filtering, such as token consumption patterns, Guardrails intervention rates and request anomalies, so security and platform teams can identify active threat campaigns and understand the full operational context of their LLM estate in production.”