Success Story
CostPatrol — AWS Cost Anomaly Detection in 24 Hours
Our own SaaS product: scans AWS accounts daily, finds savings across 30+ rules, and delivers exact fix commands with dollar amounts to Slack.
Challenge: AWS cost waste is a $30B problem. Teams get vague recommendations from Trusted Advisor or $10K/month enterprise platforms. Engineers don't have time for manual FinOps. Cost spikes go unnoticed until the monthly bill arrives.
Solution: We built CostPatrol as a serverless SaaS on the same AWS stack we use for clients. Daily scans across 30+ rules, anomaly detection within 24 hours, and specific fix commands (not suggestions) delivered straight to Slack.
Result: Published scan reports show $284/mo to $6,496/mo in savings per customer. 5-minute setup via CloudFormation. Read-only access, zero risk. Customers save 5-10x what they pay.
Tech Stack
The Story
CostPatrol came from personal frustration. I kept finding the same AWS waste on client accounts. An Aurora cluster running at $4,200/mo for a low-query search service that could have been DynamoDB plus OpenSearch at $630/mo. Metric Streams bleeding $6,900/mo because nobody had set filters, just shipping everything to New Relic across accounts and regions. CloudWatch double-shipping logs at $1,766/mo. These are real numbers from real accounts. Every client engagement started with me running the same manual checks, finding the same patterns, and thinking there has to be a better way.
So I built CostPatrol. 30+ detection rules across 8 AWS services: Lambda, EC2, RDS, DynamoDB, S3, EBS, NAT Gateway, and CloudWatch. Each rule scans real resource configurations and outputs a specific fix, not a suggestion, an actual CLI command or Terraform snippet with the dollar amount attached. The scanning engine runs daily via EventBridge, pulls data from Cost Explorer, compares to 30-day rolling averages, and flags anything that looks wrong. Anomaly detection within 24 hours instead of waiting for the month-end bill.
I built CostPatrol using Claude Code and Cursor. Specs first, AI implements against the spec, human review gates on auth, money, and security. The whole platform is serverless on Lambda, DynamoDB, SQS, and EventBridge. Slack-first delivery because engineers dont live in dashboards. Stripe handles billing with tiered pricing. CloudFormation one-click onboarding gives us read-only access to customer accounts in 5 minutes, zero risk.
The war stories are what sell it. That Aurora cluster at $4,200/mo was sitting there because "it works, dont touch it." The Metric Streams at $6,900/mo had been running unfiltered for months across multiple accounts. A single NAT Gateway burning $2K/mo because a chatty Lambda was routing through it every 30 seconds. CostPatrol finds these patterns automatically. Published scans show $284/mo to $6,496/mo in savings per customer. The goal is simple: teams spending $5K to $200K/mo on AWS should save 5-10x what they pay for CostPatrol.
How We Delivered
Our Delivery Process
See how our senior engineering pod delivered production-ready results
Daily Cost Scanning Engine
- EventBridge triggers daily scans across all connected AWS accounts
- Cost Explorer API pulls yesterday's spend, compares to 30-day rolling average
- Flags spikes >20% or >$500 absolute increase with severity and resource ARN
30+ Optimization Rules
- Covers EC2, Lambda, RDS, DynamoDB, S3, EBS, NAT Gateway, CloudWatch
- Each rule outputs specific fix: CLI command or Terraform code with dollar savings
- Rules ranked by impact across multi-account environments
Slack-First Delivery
- Anomaly alerts with fix commands delivered within 24 hours
- Daily optimization digests ranked by savings amount
- No dashboard required. Engineers get fixes where they already work.
Final Outcomes
Results
Working on something similar?
Book a 15-minute call. We'll tell you honestly if we're the right fit.
Book a 15-min Call