Case Study
AWS Cost and Compute Optimization in FinTech using AI strands agents on Bedrock
Solution
Cloud Cost Optimization | FinOps Automation |
AWS Infrastructure Monitoring
Industry
FinTech | Financial Services
Core Technology
AWS Bedrock | Agentic AI |
Claude Sonnet 4 | AWS Lambda
Overview
AWS has 200+ services that cater to each and every aspect of business optimization using the cloud.
But managing different services that you use is often difficult and costs businesses unnecessary spending.
A leading FinTech service provider was struggling with this same issue. Their AWS costs were spiraling out of control across 8 different accounts.
Xavor was called in to automate the client’s manual FinOps processes using AI agents. The autonomous, AI-driven system that we built monitored their AWS infrastructure and delivered actionable savings insights for better cloud cost optimization.
Business Challenge
Managing AWS costs across 8 accounts manually was costing cloud engineers 3+ hours every week. The reason for this bottleneck was their slow, error-prone process. Engineers had to export data to identify possible cost issues and write reports by hand.
And because reporting happened weekly, the team often discovered issues too late. They had no reliable way to detect anomalies or underutilized infrastructure across all accounts in time to act, which caused problems like:
- Incomplete coverage: Only 3 of 8 accounts analyzed due to time constraints
- Delayed insights: Weekly cadence meant missed optimization windows
- No trend detection: Hard to spot anomalies across multiple data sources
the solution
Our goal was to create intelligent AI agents that could analyze cost and utilization data. So, we built two specialized AI agents using the Strands Agent Framework and Amazon Bedrock (Claude Sonnet 4) that run on AWS Lambda, triggered by EventBridge on a schedule.
Each agent receives a plain-English prompt and independently decides which AWS tools to call. Then it pulls and processes the raw data to deliver a formatted email report without any human intervention. The entire workflow runs in under 5 minutes.
outcomes & benefits
The impact was immediate and far-reaching. Automating AWS cost management with AI agents delivered results across time and cost savings.
Time Saved :
3 hours saved every week for engineers to work on core issues.
Full Account Coverage:
100% cloud visibility as monitoring expanded from 3 to all 8 AWS accounts.
More Frequent Insights:
Reporting shifted from weekly to daily cost analysis plus weekly compute reviews.
Real Cost Savings:
$84.56/month in actionable savings due to identified EC2 right-sizing recommendations.
156 Hours Saved Annually :
156 hours saved annually, resulting in full four weeks of engineering time returned to the team every year.
Tools & tech stack
Scheduler EventBridge Scheduler
Compute AWS Lambda (Python 3.12)
AI Framework Strands Agent SDK
IaC Terraform
AI Model Amazon Bedrock, Claude Sonnet 4
Data Sources Cost Explorer API, Compute Optimizer, S3
Email Amazon SES (xavor.com domain)
conclusion
Moving to the cloud is an excellent option for any business looking to reduce costs and optimize its workflows. But it is very easy to bleed money in the cloud without proper visibility in your cloud environment.
Xavor helped this client go from reactive, incomplete cost tracking to a fully automated, intelligent monitoring system using AI agents to achieve full visibility and surfacing real savings from day one. More importantly, it gave the client a scalable foundation for proactive cloud cost governance, future predictive budgeting, and automated remediation.
Optimize your cloud environment now
Ensure your cloud resources are delivering maximum results