Case study, AI agents · Enterprise IT operations
An AI agent that plans server patching, with a human approving every change.
A GenAI workflow on Amazon Bedrock that works with patch data and the IT change process, with human approval steps built in.
The problem
Monthly patching across large server estates was slow and manual: checking compliance, planning maintenance windows, writing change tickets and chasing approvals.
How it's built
- Amazon Bedrock LLM agent that reasons over patch compliance data from AWS Systems Manager Patch Manager.
- Integrated with the client's ITSM process so patch work flows through normal change management.
- Human-in-the-loop approval step before any patch is applied.
- Least-privilege IAM roles scoped to exactly what the agent may touch.
- Python orchestration with audit logging.
What shipped
A secure GenAI operations agent integrated with the client's ITSM process, with guardrails and audit trails built in.
Clean handoff
Designed around least-privilege access, human approval and audit trails, so the client stays in control.
Talk about a build like this →