Greenlight Financial Luna security operations agent
Greenlight Financial uses Luna to investigate security and financial-risk alerts, resolve routine cases, and escalate evidence to analysts.
Reported results
- More than 90% autonomous resolution rate
- 3,000-4,000 cases processed per month
- Routine median time-to-triage fell from 15-30 minutes to seconds
- Estimated 900-1,800 analyst hours avoided per month
As stated by AWS. Not independently verified.
- Company
- Greenlight Financial
- Agent
- Luna
- Job
- Security
- Industry
- Financial services
- Platform
- AWS Bedrock
- Type
- Company deployment
- Source
- AWS (aws.amazon.com)
- Added
- Nov 11, 2026
What the agent does
Greenlight Financial's Luna is a tiered multi-agent system that triages breach-card, identity, infrastructure, and financial-risk events. More than 30 specialist agents query systems such as Tines, DynamoDB, Mixpanel, and Zendesk, while AgentCore memory helps recognize recent patterns and prior resolutions. Routine cases are resolved automatically; complex or high-risk cases move to deeper investigation and then to human analysts with an evidence chain. AWS reports more than 90% autonomous resolution across 3,000-4,000 monthly cases.
Read the source on aws.amazon.comStack
Related cases
Duco financial operations agents
Duco runs tenant-isolated agents to onboard reconciliation processes and prepare financial-data exceptions for analyst decisions.
Sumo Logic Mobot security analysis agent
Sumo Logic runs Mobot and specialist agents to interpret security telemetry, combine alerts, and suggest investigation steps.
CTBC Bank COBOL reverse-engineering agents
CTBC Bank uses a governed multi-agent pipeline to recover business rules from COBOL and generate specifications, SQL, and tests.
Trovata treasury operations agents
Trovata runs autonomous treasury agents that monitor cash activity, reconcile accounts, explain variances, and prepare reports.
Nubank disaster recovery evidence agent
Nubank's agent gathers recovery evidence from metrics, logs, and alerts and produces audit-ready reports for disaster recovery tests.
Abnormal AI email threat investigation agents
Abnormal AI agents investigate difficult email threats with sandboxed code and draft detection heuristics from misclassification patterns.