Rubrik boosts Amazon Bedrock security with AI agent monitoring
Rubrik has unveiled Rubrik Agent Cloud for Amazon Bedrock AgentCore, targeting security and governance for organisations deploying AI agents across multi-cloud platforms.
The new integration is set to provide Amazon Bedrock customers the ability to monitor, audit, manage, and remediate actions by AI agents, giving security and IT teams more control over automation within their environments.
AI agent lifecycle
The integration promises to deliver a unified approach for managing the full lifecycle of AI agents. Enterprises deploying agents on Amazon Bedrock AgentCore will be able to view active agents, monitor their actions, set policies in real-time, and reverse unintended changes efficiently.
The solution aims to address growing concerns around the operational and security risks posed by expanding use of agentic AI.
"Amazon Bedrock AgentCore gives developers the flexibility to build and deploy intelligent agents using their choice of AI model or framework. With Rubrik Agent Cloud, organisations will have a single plane of glass for the complete AI agent lifecycle, one that enables customers to see every agent in action, enforce policies in real-time, and instantly rewind mistakes before they become catastrophic errors," said Devvret Rishi, General Manager of AI, Rubrik.
Operational risk
Organisations are experiencing increased cyber risks associated with the use of generative AI.
Research cited by Rubrik notes that 72% of business leaders report an upswing in cyber threats, with nearly half attributing this to AI-driven capabilities being used in attacks. Rubrik Agent Cloud is positioned as a control layer that supports the mitigation of these risks, providing automated oversight for agent activities and enabling security policies to be applied centrally.
The integration enables discovery of both standard and custom agents built on Amazon Bedrock AgentCore. Security teams gain continuous monitoring of agent actions and comprehensive audit trails. Policies around access, agent behaviour, and allowed actions can be centrally enforced, with connectivity to enterprise identity systems for compliance and governance requirements.
Error rollback
The integration includes a remediation capability known as Agent Rewind, which offers precise, time-based rollback of unwanted or destructive agent actions. This selective rollback is designed to minimise downtime and data loss while preserving workflow continuity and system integrity. Organisations can recover from agent-driven mistakes without pausing broader operations.
Rubrik Agent Cloud for Amazon Bedrock AgentCore is structured to cover the spectrum from monitoring to performance management, with simulation capabilities also available. The company states not all features are immediately available but will be rolled out over time.
Resilience recognition
Rubrik also announced it has achieved the AWS Resilience Software Competency, specifically in the recovery category. This recognition is intended for AWS partners offering validated systems that enhance critical application availability and resilience, with solutions tailored to specific customer needs.
Amazon Bedrock AgentCore's growing popularity among developers is seen as increasing the need for governance frameworks. Rubrik's move addresses this demand by focusing on the security and rollback requirements of enterprises venturing into automation using large-scale AI agents.
"With Rubrik Agent Cloud, organisations will have a single plane of glass for the complete AI agent lifecycle, one that enables customers to see every agent in action, enforce policies in real-time, and instantly rewind mistakes before they become catastrophic errors," said Rishi.
