Why I built an AWS cost tool for teams that don't have a FinOps person
Every AWS cost cleanup I've been part of followed the same curve. Someone spends a week auditing, kills the obvious waste, the bill drops, everyone's happy. Six months later it's back where it started, and nobody can say exactly when.
It's not that the cleanup was wrong. It's that a cleanup is a one-time action against a system that changes every week. A load test cluster stays up over a weekend. An engineer leaves and their EBS volumes don't. A NAT gateway quietly becomes a top-three line item because someone routed traffic through it and never looked again.
The tools that solve this properly assume you have a person whose job is cloud cost. Enterprise FinOps platforms are built around a rollout, a dashboard with an owner, and pricing that only makes sense at that scale. Startups don't have that person. They have one backend engineer who notices the bill went up and loses an afternoon in Cost Explorer trying to find out why.
So Parsivex is deliberately small:
Connect a read-only IAM role (CloudFormation or Terraform, ~10 minutes)
Run a free scan -- idle EC2, unattached EBS, oversized RDS, stale snapshots, NAT gateway overuse, and a few dozen other patterns
Then the part that actually matters: scheduled re-scans comparing findings month over month, and a daily anomaly check against a rolling baseline, with alerts to email or Slack
The scan is the hook. The monitoring is the product. I'm not going to claim it pays for itself, I don't know your bill. What I'll claim is that finding out about drift on a Tuesday morning is meaningfully better than finding out on next month's invoice.
Honest about the friction: you do need to sign up and set up the IAM role. There's no instant demo. It's AWS-only, on purpose — I'd rather do one cloud properly than three badly. And it isn't a replacement for a real FinOps platform if you're at the scale that needs one.
Full disclosure, I'm the founder. The scan itself is free, no credit card: https://www.parsivex.com/?utm_source=indie_hackers&utm_medium=referral&utm_campaign=evergreen
For IH folks: PM me and I'll send you a code for a free month of Monitoring — scheduled re-scans and daily anomaly alerts, the paid tier. I'd rather have ten people actually running it for a month and telling me what's wrong with it than a hundred one-off scans. No catch, and I'm not going to chase you at the end of it.
Two things I'd genuinely like input on:
1. For those of you running on AWS — what's the line item that surprised you most when you finally looked?
2. Is a daily anomaly alert useful, or just another notification you'd mute in week two? That's the design question I keep going back and forth on.
Happy to walk anyone through the IAM setup or explain how a specific finding type is detected.
great writeup. the two questions you asked at the end are exactly
what every small aws team struggles with. for us the most surprising
line item was never the big instances, it was the quiet stuff: a
detached ebs volume, a staging box left running, a lambda version
nobody remembers. on anomaly alerts, the early ones get muted fast
when they fire on noise, so the ones that stick are tuned to a real
baseline, not a flat threshold. curious what you settled on.
AWS cost optimization has two buyers with opposite needs. The dev who wants to clean up their $50/mo side project account wants a one-time scan and a "delete this" button. The FinOps team managing $50K/mo wants continuous monitoring, Slack alerts, SSO, and compliance reporting. "Scan for waste" attracts the first. "Keep watching" hints at the second — but your $0/mo pricing and lack of enterprise features signal "this is for personal accounts, not production infrastructure.
The “scan is the hook, monitoring is the product” distinction is strong.
On the daily alerts, have you seen any evidence yet about what users actually do after receiving one — investigate immediately, come back later, or mostly ignore it? Feels like that behavior matters more than whether they say they want daily alerts.