How Tamnoon built AI-powered cloud security remediation with Amazon Bedrock
aws ·
Tamnoon has developed Tami, an agentic cloud security remediation orchestrator built on Amazon Bedrock, to address the overwhelming volume of alerts generated by CNAPPs and posture scanners. Tami autonomously investigates findings from multiple cloud providers, assesses their safety using a specialized scoring agent, and generates validated fixes with strict guardrails against production impact. The architecture leverages four distinct AI agents running on AWS Lambda and EKS, utilizing Bedrock’s ability to route tasks to the most appropriate foundation models for reasoning, validation, or code generation. Security is maintained through data isolation within the customer's AWS environment, prompt injection protection via Bedrock Guardrails, and least-privilege IAM permissions. By clustering root causes and prioritizing SAFE actions, the platform achieved a 90% reduction in critical cloud exposure for customers within 90 days. Notably, this automation resulted in zero production issues while saving over 13,000 engineer hours per week across its user base. The solution integrates seamlessly with existing tools like Wiz and AWS Security Hub, normalizing findings into a unified schema for consistent remediation. This approach demonstrates how agentic AI can safely scale security operations without requiring proportional increases in human headcount.
aws