How Greenlight Financial built an autonomous SOC solution on Amazon Bedrock AgentCore

Judgment

Greenlight’s SOC and risk leaders should decide whether to lift Luna’s autonomy above the stated more than 90 percent, accepting residual human escalation and Amazon Bedrock AgentCore dependency as the cost of less manual triage.

Basis

In an AWS Machine Learning Blog post dated 06 NOV 2026, Adam Maksimuk, Gaurav Begwani, Mani Teja Nissankararao, and Venu Thangalapally write that Greenlight Financial resolves more than 90 percent of up to 4,000 security and risk cases each month without human intervention. Luna uses Amazon Bedrock AgentCore runtime to host agent code and AgentCore memory for a tiered multi-agent system, avoiding building and operating those capabilities in-house. Luna triages dark web breach-card alerts, identity anomalies, and financial-risk events in seconds; the architecture includes a multi-tiered supervisor, more than 30 specialized agents, episodic memory for continuous learning, and expansion from SOC into financial risk triage. The mechanism is agent orchestration plus managed runtime and memory; the decision hinges on how much residual escalation and vendor dependency Greenlight will accept at the more than 90 percent line.