Crego.ai

Overview

Crego — India's fastest-growing AI-native lending platform — faced a critical challenge weeks before its launch. Legacy on-premise MySQL and rule-based workflows would collapse under the scale of thousands of real-time AI credit decisions per minute.

The Challenge

• Legacy monolithic lending systems with manual credit evaluation.
• Slow rule-based processes: ~2–3 days per loan application.
• No AI-driven decisioning or workflow automation.
• Security gaps: no encryption at rest for sensitive customer data.
• Zero ability to scale to multi-tenant, high-throughput workloads.

AWS AI Platform Built

• Amazon Bedrock (Claude-2) for semantic analysis of loan applications.
• Amazon ECS + Fargate for AI workflow automation and real-time decisioning.
• Amazon RDS MySQL for structured transactional data.
• Amazon S3 for unstructured loan documents.
• Auto-scaling ECS clusters for thousands of concurrent AI scoring requests.

Launch Day Results

• 4,300+ concurrent active loan applicants.
• 180,000+ AI decisions processed per hour.
• p95 query latency < 20ms; end-to-end offer generation < 90ms.
• 99.999% uptime including Multi-AZ failover.
• Zero disruption to customers — real-time approvals and instant satisfaction.

Business Impact

• Credit evaluation time reduced from days to minutes (85% faster).
• Operational cost reduction of ~40% via AI automation.
• 50% reduction in total cost of operations vs legacy systems.
• Multi-tenant AI workloads scaled across multiple financial institutions.