DahReply

 

DahReply, Malaysia’s AI chatbot provider,

About Us

partnered with Agmo to modernize its conversational AI platform on AWS using DevOps best practices. Automated CI/CD, monitoring, and container orchestration imp

roved scalability, resilience, and operational efficiency.

Customer Background & Problem Statement

DahReply delivers AI-powered chatbots for SMEs and enterprises across Asia.

 

Challenges:

    • Manual deployments causing downtime and long release cycles.
    • Lack of observability across chatbot workloads.
    • Difficulty scaling across multiple digital channels.
    • Increasing cost and security concerns with growing customer data.

Proposed Solution & Architecture

Agmo implemented a DevOps-driven cloud-native architecture on AWS:

  • Infrastructure as Code (IaC): Terraform templates for consistent provisioning across environments.
  • CI/CD Pipeline: GitHub Actions integrated with Amazon ECR & Amazon EKS for blue/green and rolling deployments.
  • Container Orchestration: Amazon EKS with auto-scaling and service mesh for AI inference workloads.
  • Monitoring & Logging: Amazon CloudWatch, Prometheus, and Amazon OpenSearch for observability.
  • Security & Compliance: IAM roles with least privilege, AWS KMS for secrets, and automated image scanning.
  • GenAI Enablement: Amazon Bedrock integrated with EKS to serve domain-specific chatbots.

Outcomes & Success Metrics

  • Reduced deployment time from days to <30 minutes with CI/CD automation.
  • Improved system reliability with 99.9% uptime and automated rollback.
  • Enhanced scalability, supporting 5x more concurrent users without downtime.
  • Increased release velocity — weekly releases instead of quarterly.
  • Achieved cost savings of ~30% by adopting AWS managed services.

TCO Analysis

  • Eliminated CapEx with cloud-native deployment.
  • Reduced OpEx through automated scaling and managed observability tools.
  • Lowered DevOps team effort by ~40% due to automation and IaC.

Lessons Learned

  • Early investment in IaC and CI/CD pipelines accelerated innovation.
  • Shift-left security with image scanning and secrets management reduced vulnerabilities.
  • Observability (logs, metrics, traces) was critical for AI workload reliability.
  • Partnering with AWS enabled DahReply to confidently scale its GenAI workloads.