I built this because I was tired of spending 20-30 minutes investigating every CloudWatch alarm that fired. AWS just released CloudWatch Investigations (June 2025) which does similar things, but requires significant setup and lives in the console. Mine takes a different approach.
CloudWatch AI Agent: When an alarm fires, an AI agent autonomously investigates and sends findings to Slack. Deploy via Terraform in 5 minutes.
How it works: - Terraform module (Apache 2.0) deploys Lambda + SNS integration - Alarm triggers → AI agent queries your AWS environment (read-only) - Analyzes metrics, logs, EC2/RDS/Lambda configs, alarm history - Sends root cause analysis + ready-to-run CLI commands to Slack - ~30 seconds total
vs AWS CloudWatch Investigations: - CloudWatch Investigations: Console-based, requires investigation groups, IAM role setup, CloudTrail→CloudWatch Logs config, limited to 2 concurrent investigations - Mine: Terraform apply, works immediately, Slack-native, unlimited investigations
The Terraform module is fully open source. The Lambda code is obfuscated (contains AI orchestration logic, prompts) - this is the core IP. Business model is $5/mo API key subscription.
Cost: $5/mo + ~$0.001 per alarm (AWS Bedrock usage)
Why build this when AWS has a free native solution? 1. Infrastructure-as-code teams want Terraform deployment, not console configuration 2. Slack-native workflow vs AWS console workflow 3. No investigation limits (AWS caps at 150 enhanced/month) 4. 5 minutes vs 2-3 hours of setup
I'm a solo AWS engineer. Built this for myself, now productizing it. The obfuscated Lambda is a tradeoff - I know trust is an issue, but the Terraform is fully auditable, IAM permissions are read-only, and everything runs in your account with CloudTrail logging.
Not for everyone (especially teams with strict "no obfuscated code" policies), but for Terraform-first DevOps teams who want simple tooling, it solves a real pain point.
Site: https://aiopscrew.com
Happy to answer questions about the implementation, security model, or why I made specific design decisions.