Greenlight Financial says its AI system, Luna, resolves more than 90% of its 3,000–4,000 monthly security and risk cases without human intervention. Built on Amazon Bedrock AgentCore, Luna routes routine cases through more than 30 specialized agents and sends harder investigations to a second-level system or human analysts.
The company says routine cases take a median of seconds to triage, compared with 15–30 minutes for manual review. Greenlight estimates the system avoids 900–1,800 hours of routine analyst work each month. It expanded Luna from security operations into financial-risk triage within two months; the results are Greenlight’s own reported figures.
Why it matters
For Greenlight’s security analysts, the reported automation could free up hundreds of hours each month for cases needing human judgment, though the figures come from the company and do not establish how the system performs elsewhere.