How Tamnoon built AI-powered cloud security remediation with Amazon Bedrock
Tamnoon built Tami, an agentic cloud remediation orchestrator on Amazon Bedrock that autonomously investigates, prioritizes, and safely remediates cloud‑security findings. The system reduces critical cloud exposure by over 90 percent within 90 days while maintaining zero production incidents. It does this with four specialized AI agents running in an event‑driven, serverless AWS architecture that scales each remediation stage independently. Engineers can leverage this pattern to automate high‑volume security triage without sacrificing safety or governance.
⚡ Key Takeaways
- Over 90 percent reduction in critical cloud exposure achieved in 90 days.
- Four specialized AI agents (investigation, prioritization, safety assessment, and fix orchestration) each handle a distinct remediation phase.
- Event‑driven, serverless architecture on AWS enables independent scaling and safe failure of each agent.
- Integration point: findings are ingested from CNAPP, CSPM, and other security tools and normalized into a unified schema for the agents.
- Guardrails and governance are enforced before any change touches a live environment, ensuring production‑safe fixes.
🔧 Tools & Libraries
By decomposing remediation into isolated, serverless AI agents on Bedrock, teams can automate the bulk of security triage at scale while preserving the safety guarantees required for production cloud environments.
✅ Practical Steps
- Ingest security findings into Tami’s unified schema from your CNAPP or CSPM tools as described.
Want the full story? Read the original article.
Read on AWS ML Blog ↗