Stop GenAI Account Takeover Fraud Using AWS Automation

Accelerating Fraud Resolution with GenAI

A leading fintech company faced rising account takeover fraud requests across apps, helpline, chat support, and email, impacting customer satisfaction and operational efficiency. Tech Mahindra partnered with AWS to enhance fraud investigation through a real-time fraud analyst dashboard and a GenAI-powered conversational interface, enabling analysts to investigate suspicious transactions using simple business language

AWSFintech-thumb

Business Challenge

The customer, a leading Philippines-based fintech company specializing in mobile payments and remittance, is a crucial player in the country’s digital finance ecosystem. The customer experience (CX) team of the company repeatedly raised various issues regarding account takeover (ATO) and frauds like:

  • Receiving ATO fraud requests through apps, helpline, chat support, and email
  • Slow resolution due to manual processes that increased analyst workload
  • Hinderance in accurate fraud detection and resolution posed by missing or incorrect data
  • The customer expected the resolution time to reduce from 5-days to a 1-day

Our Approach and Solution

Tech Mahindra partnered with AWS to design a scalable, AI-enabled fraud investigation solution powered by data engineering, automation, and GenAI-driven analysis.

AI-led Fraud Investigation

We developed a GenAI-powered conversational interface that allows fraud analysts to ask questions and investigate suspicious ATO transactions using simple business language.

Real-time Fraud Monitoring

The fraud analyst dashboard monitors transaction dispositions in real time and flags suspicious account takeover transactions for investigation.

AWS-based Automation

The future-state EI Anti-Fraud Solution used AWS Step Functions to automate calculations and AWS Bedrock to orchestrate the conversational interface for analysts.

Scalable Cloud Architecture

The solution integrated AWS services including EKS, ECR, Load Balancer, PostgreSQL, OpenSearch, DynamoDB, Lambda, S3, EFS, Bedrock, and Amplify.

Improved CX Enablement

A chatbot-supported experience helped CX agents respond to fraud-related queries faster and more efficiently.

Business and Community Impact

  • ATO disposition flagging reduced from 5–7 days to under 15 minutes
  • Near real-time investigations within approximately 5 minutes of report submission
  • Automated validation across 30 success criteria, eliminating manual checks
  • 10%+ CSAT uplift recorded in customer satisfaction
  • Enhanced fraud team decision-making through conversational access to dispute and ATO data.