A major regional insurance provider leveraged Amazon Nova models on Amazon Bedrock to deliver policy-aware AI assistants that automate routine enquiries, improving customer satisfaction and reducing operational costs.

The Challenge

The insurer faced growing challenges in delivering clear, efficient, and compliant customer service.

• Modernise how customers access and understand complex policy information.
• Reduce the pressure and workload on call centre teams.
• Maintain strict compliance with industry regulations while scaling service.

Our Solution

Rambunct implemented a secure, serverless AI solution on AWS to deliver policy-aware, human-like customer interactions at scale.

• Deployed Amazon Nova models on Amazon Bedrock to build conversational AI assistants that explain policy documents in plain language and provide personalised recommendations.
• Used AWS Lambda to securely fetch real-time customer policy data from backend systems, enriching each interaction.
• Delivered the solution via serverless Bedrock API calls, ensuring scalability without in-house AI infrastructure.
• Integrated within a secure VPC environment with AWS KMS-encrypted data handling and Amazon CloudWatch telemetry for monitoring and compliance.

The Result

The solution delivered significant efficiency gains, cost savings, and improved customer experiences.

• 60% of routine queries fully automated.
• 35% reduction in operational costs over three years.
• 40% faster customer response times.
• 17% improvement in customer satisfaction ratings.

The post Modernising Insurance Customer Service with Amazon Bedrock AI appeared first on Rambunct Consulting.

Rabattcorner, a cashback platform, needed an intuitive analytics layer to visualise Daily, Weekly, and Monthly performance across Gross Merchandise Value (GMV), Revenue, Payments, Average Order Value, Number of Transactions, Transaction Statuses, conversion rate, and more. The team had been using Looker, but sought the advantages of Amazon QuickSight: built-in machine learning insights (anomaly detection, forecasting, automated narratives), a simpler interface with natural-language query (NLQ) for non-technical users, high performance with the SPICE in-memory engine, and a cost-efficient pay-per-session model.

The Challenge

Rabattcorner set out to replace Looker with Amazon QuickSight while improving usability, performance, coverage of key KPIs, and overall cost efficiency.

• Built-in ML insights: Achieve anomaly detection, forecasting, and automated narratives without stitching in external tools.
• Ease of use for business users: Reduce reliance on LookML and enable ad-hoc analysis via natural-language query (NLQ).
• Performance at scale: Speed up interactive dashboards and keep response times consistent on large datasets.
• Cost control: Move from a fixed subscription to QuickSight’s pay-per-session model aligned to actual usage.
• Comprehensive KPI coverage: Track Daily/Weekly/Monthly GMV, Revenue, Payments, AOV, Transactions, Statuses, and conversion in one place.

Our Solution

We implemented a QuickSight-based analytics stack and rollout plan tailored for fast performance and broad self-service adoption.

• Role-based dashboard suite: Built Daily/Weekly/Monthly views for GMV, Revenue, Payments, AOV, Transactions, Statuses, and conversion.
• SPICE acceleration: Modelled and loaded data into the in-memory SPICE engine for fast, interactive exploration.
• ML-powered analytics: Enabled built-in anomaly detection, forecasting, and auto-generated narratives.
• Self-service NLQ: Activated natural-language queries so business teams could ask questions directly.
• Data validation & metric parity: Structured verification to ensure complex calculated fields matched legacy definitions.
• Adoption & enablement: Phased pilot with power users, organisation-specific documentation, and recorded training sessions.
• Performance governance: Established baselines, monitoring, and SPICE-capacity management to keep dashboards responsive.

The Result

The migration delivered measurable gains in speed, cost, and adoption, validated by a three-year TCO analysis.

• Faster delivery: Report development time dropped from 2–3 weeks (Looker) to 3–5 days (QuickSight) (~70% faster).
• Snappier dashboards: Load times improved from 5–7s to 2–3s, boosting usability and adoption.
• Lower support effort: Ongoing effort reduced from 20 hrs/month to 8 hrs/month (60% reduction).
• Higher adoption: Active business users increased from ~40% to ~75% thanks to NLQ and simpler workflows.
• Built-in ML value: Anomaly detection, forecasting, and narratives delivered out of the box, with no external tools required.
• TCO savings: Three-year total fell from $336,000 (Looker) to $167,600 (QuickSight): $168,400 saved (~50%), with ~40% lower licensing and 60% lower ops costs.
• ROI & payback: >300% ROI in Year 1, payback in <3 months, and ~$56,000/year in ongoing licensing and labour savings.

The post Rabattcorner: Cashback Platform Streamlines Analytics with Amazon QuickSight appeared first on Rambunct Consulting.

The Challenge

• Outdated Django stack and Python version leading to security and compatibility issues
• Deprecated MySQL database causing instability and performance problems
• Lack of CI/CD pipelines and standardised testing increased deployment risks
• Manual scaling and infrastructure drift added operational overhead
• Inefficient tooling and workflows reduced developer productivity

Our Solution

• Migration to Python 3.11, Django 4.2.14, and MySQL 8
• Deployment on AWS Graviton2/Graviton3-based Amazon EC2 instances
• Use of Application Load Balancer (ALB) and Auto Scaling groups
• End-to-end automation and security using GitHub Actions, AWS CodeDeploy, AWS CloudFormation, AWS WAF, AWS Secrets Manager, Amazon Inspector, and AWS Certificate Manager
• Optimised performance and observability with Amazon RDS, Amazon S3, Amazon ElastiCache, Amazon CloudWatch, and AWS X-Ray

The Result

The Rambunct architecture enabled blue-green deployments, secure secrets management, observability, and scalable, resilient infrastructure.

• 35% improvement in response time due to Python and Django upgrades
• 40% reduction in compute costs by leveraging AWS Graviton processors
• 90% faster deployment cycles via automated CI/CD workflows
• 95% code coverage achieved with integrated test automation
• Improved performance and reliability with 60% faster MySQL queries, 99.9% uptime, and zero-downtime deployments via a blue-green strategy
• Significant operational efficiency with a 75% reduction in manual infrastructure tasks through automation and autoscaling

The post Nitorum: Django Application Modernisation on Amazon EC2 Case Study appeared first on Rambunct Consulting.

We implemented a highly secure, containerised insurance platform using Amazon ECS with multi-zone replication, automated scaling, and encrypted data handling for real-time insurance operations.

The Challenge

• Secure & Isolated Containers: Implemented ECS with private subnet isolation, encrypted data handling, and integration with AWS KMS and existing security frameworks.
• High Availability & Scalability: Enabled multi-zone replication, CPU-based auto-scaling, and regional failover to support dynamic workloads and resilience.
• Robust Service Communication: Deployed advanced service discovery using private DNS namespaces to manage complex internal service interactions efficiently.

Our Solution

• Secure, Resilient Infrastructure: Deployed ECS services in private subnets with AWS WAF, KMS encryption, certificate verification, secret management, and multi-AZ failover for security and high availability.
• Modular ECS Architecture: Used ECS Service Groups for frontend, backend, API gateway, and Redis cache, with service discovery, private DNS, CPU-based auto-scaling, and integrated load balancing.
• Efficient Data & Lifecycle Management: Managed structured and unstructured data via Amazon RDS and Amazon S3; implemented encrypted transmission, caching with Redis, and active lifecycle flows from frontend to storage.

The Result

The solution delivered high performance, operational efficiency, and robust security through a well-architected AWS containerised setup.

• High Availability & Speed: Achieved 99.99% uptime and sub-100ms service discovery for responsive and reliable service delivery.
• Scalable Infrastructure: Enabled successful auto-scaling events to handle dynamic workloads without service disruption.
• Operational Efficiency: Reduced deployment time, improved resource utilisation, and simplified system management.
• Strengthened Security: Maintained a zero-breach record while enhancing the overall security posture through encryption and isolation.

The post Secure Multi-Zone Amazon ECS Deployment for Modern Insurance Operations appeared first on Rambunct Consulting.