The promise of AI agents, boosting productivity and automating complex tasks, is undeniable. However, unchecked autonomy can lead to real-world consequences, especially when agents interact with sensitive data or critical systems. Research indicates a significant rise in cyber risks linked to generative AI, making robust control measures essential.
Rubrik Agent Cloud directly addresses this need, providing security and IT teams with a control layer to manage the rapidly expanding attack surface created by AI agents. This integration with Amazon Bedrock AgentCore aims to provide peace of mind as organizations increasingly rely on AI.
Rubrik Agent Cloud employs a comprehensive strategy encompassing monitoring, governance, and remediation to ensure AI agents operate within defined boundaries and don’t inadvertently cause harm.
Agent Monitor: Eyes on the AI
The monitoring component automatically discovers agents built on Amazon Bedrock AgentCore, as well as custom agents, and maps their potential impact across various environments. It continuously tracks agent activity and data access, maintaining immutable audit trails with rich context.
- Auto-discovers agents built on Amazon Bedrock Agentcore, as well as custom agents.
- Automatically discovers and maps active agents’ blast radius across Amazon Bedrock and other popular agent building tools.
- Continuously monitors agent activity and data access, and maintains immutable audit trails capturing context from data, identity, and applications.
Agent Govern: Setting the Rules of Engagement
Governance features allow teams to track agent usage, evaluate performance against prompts, and control undesirable actions. By defining and enforcing behavior policies in real-time, Rubrik helps ensure agents adhere to established security and compliance standards. This connectivity with enterprise identity systems further strengthens control and accountability.
- Tracks agent usage, evaluates performance against prompts, and gives teams the tools to control undesired actions.
- Defines and enforces agent behavior, access, and action policies in real-time.
- A centralized tool to provide connectivity with enterprise identity systems—helping ensure secure, compliant, and controlled innovation.
Agent Remediate: The AI “Undo” Button
Perhaps the most intriguing aspect is Agent Remediate, featuring “Agent Rewind,” which integrates with Rubrik Security Cloud. This functionality allows for precise rollback of undesirable or destructive actions, effectively providing an “undo” button for AI mistakes. This goes beyond simple observability, allowing for immediate correction without downtime or data loss.
- Announced in August 2025, Agent Rewind integrates with Rubrik Security Cloud to provide the industry’s only solution for precise time and blast radius rollback of undesirable or destructive actions.
- Goes beyond observability to allow customers to instantly undo unwanted or destructive actions, without downtime or data loss.
- Selective rollback of agent-driven changes ensures continuous protection for critical data and systems, and immutable recovery.
Adding another feather to its cap, Rubrik also announced it has achieved the AWS Resilience Software Competency in the Recovery category. This recognition validates Rubrik’s ability to help customers improve the availability and resilience of their critical systems using AWS Resilience Services.
Rubrik Agent Cloud for Amazon Bedrock AgentCore promises to be a crucial tool for organizations navigating the complexities of AI adoption. By providing comprehensive monitoring, governance, and remediation capabilities, Rubrik is helping to ensure that AI agents are not just powerful, but also responsible and secure. As AI continues to permeate every facet of business, solutions like this will be paramount in mitigating risks and unlocking the true potential of intelligent automation.
