AWS cost optimization for Slack

One idle m5.xlarge costs you $140/month

SaneCost watches your AWS infrastructure and pings you in Slack the moment an instance goes idle. Setup takes 5 minutes.

Add to Slack — Free How it works

No credit card / 5-minute setup / Read-only AWS access / Your thresholds, your rules

# aws-alerts

SaneCost 2:34 PM

Idle Instance Detected

i-0a4f2e8b1c in us-east-1 has been below 2% CPU for 6 hours.

Type m5.xlarge Cost $140/mo

SaneCost 2:34 PM

Owner @alice notified via DM.

Example Slack notification from SaneCost

You're not alone

If your AWS bill feels wrong, it probably is. The industry wastes $44.5 billion a year on idle cloud infrastructure. Most teams don't even know it's happening.

Don't take our word for it:

$44.5B in cloud waste projected for 2025 — Harness 32% of cloud budgets wasted on idle resources — DataStackHub 60%+ of waste from idle & overprovisioned compute — CloudZero

What it does

Catches the waste your dashboards miss

Detection

Analyzes CPU utilization patterns across your fleet to identify truly idle instances — filtering out batch jobs, periodic workloads, and freshly-launched instances that haven't ramped up yet.

Pattern analysis / Owner extraction from tags / False positive filtering / Configurable thresholds

$ sanecost analyze i-0a4f2e8b1c

Instance i-0a4f2e8b1c · m5.xlarge · us-east-1a

CPU avg 1.2% over 6h · Network 0.3 MB/h · Threshold 5% (configured)

Owner alice@company.com (from tags)

▶ Verdict: idle — $140/mo recoverable

Your rules, not ours

You define what "idle" means for your infrastructure. Set your own CPU thresholds, observation windows, and instance exclusions. Every team's workload is different — SaneCost adapts to yours, not the other way around.

Slack Alerts

Posts to your team channel with full context — instance ID, type, region, daily cost, and how long it's been idle. If the instance has an Owner tag with an email, the owner gets a direct DM too.

Channel + DM delivery / Owner auto-detection / One-click stop

aws-alerts

Idle instance i-0a4f2e8b1c — $140/mo

+ owner gets a DM+

DM to @alice

You own an idle instance — tap to stop it

Monitoring

Scans across all regions in your connected account. Deduplicates alerts so you're never spammed about the same instance. State changes are tracked in real time.

Multi-region / Real-time tracking / Alert deduplication

$ sanecost status

Account 1 connected · Regions us-east-1, eu-west-1, ap-southeast-1

Last scan 2 min ago · Next in 28 min

Instances 147 monitored · Alerts 3 idle (deduplicated)

✓ Account healthy — no duplicate alerts sent

Setup

Three steps.
Five minutes.

From install to first alert in under five minutes.

Start now

Install the Slack app

One click to add SaneCost to your workspace. Choose which channel receives alerts.

Connect your AWS account

Deploy a CloudFormation template that creates a read-only cross-account role. Takes under 2 minutes.

Get your first alert

SaneCost scans immediately. Idle instances show up in Slack with full context and a stop button.

Security

Built for teams that care about access control

Read-only access

We only read CloudWatch metrics and EC2 metadata. Stop actions require your explicit click.

TLS encrypted

All data in transit is encrypted. API endpoints are secured and rate-limited.

No data stored

We don't persist your AWS data. Analysis happens in real-time, results go to Slack, nothing is retained.

IAM role-based

Standard AWS cross-account IAM role with least-privilege permissions. You control access.

Pricing

Free while we're in early access

Full access to everything. No limits, no credit card, no strings. We'll introduce paid tiers later — early adopters will be grandfathered.

What's included

One AWS account per workspace

Unlimited team members

Idle instance detection

Owner DM notifications

One-click stop from Slack

Email support

Get started free

How it works under the hood

Algorithms first, AI when it helps

Core detection is algorithmic — CPU thresholds, time windows, cost calculations. No black box. AI steps in only for edge cases: refining confidence on ambiguous patterns and extracting owners from messy tags. Every alert ships with raw data so you can see exactly why it was flagged.

You decide. We never act without your click. But we'll nudge you towards savings like nobody else can.

AI-assisted recommendations may occasionally be inaccurate. Every alert includes the data you need to decide — one glance, one click.