A walkthrough for building an AI agent that automatically detects AWS cost anomalies using Amazon Bedrock (Claude Sonnet) and Elastic Cloud Serverless. The agent reads billing data from Elasticsearch indexes populated via the Elastic agentless collector, compares current costs against a 7-day rolling average, correlates spikes with deployment events, and sends root-cause alerts to Slack. The setup uses AWS Lambda triggered by an EventBridge cron job at 8 AM daily, with secrets stored in AWS Secrets Manager. The project code was generated using Kiro (AWS AI coding tool) and is available on GitHub.
•29m watch time
85 Impressions