One engine. Every cloud cost signal.
Finopscraft ingests raw billing data from AWS, GCP, and Azure, normalises it across providers, and runs it through a time-series forecasting model trained on your specific deployment patterns. The result: a next-month forecast accurate to within 8%, plus a ranked list of waste to fix today.
ARIMA plus deployment-aware anomaly detection
Most cloud forecasting tools project last month forward in a straight line. That works when your workload is stable. It fails the week after you ship a new service, migrate a database, or run a quarterly batch job. Finopscraft's model separates the trend component from deployment-driven spikes and seasonal patterns before projecting forward.
Weekly retraining against actuals keeps forecast error in check. When error rate rises above threshold, the model flags itself and adjusts the confidence interval shown in your digest.
- Per-service forecast: EC2, RDS, S3, GCE, BigQuery, Azure VM, Blob Storage, and 12+ others
- Per-region breakdown: us-east-1, eu-west-2, and others tracked independently
- Per-team tag attribution: map spend to eng, product, data, and custom allocation tags
- 30, 90, or 180-day forecast window depending on your tier
Find the spend you can actually cut
Finopscraft checks your infrastructure against a set of waste rules updated with each provider's pricing model. Three categories account for the majority of cloud waste we find in real environments.
- Idle compute: EC2, GCE, and Azure VMs with CPU below 5% for 7+ consecutive days
- Overprovisioned storage: RDS and Cloud SQL instances sized for peak that now run below 15% capacity
- Orphaned resources: EBS volumes, GCP persistent disks, snapshots, and load balancers with no attached workload
Each flagged resource includes a specific action: terminate, resize, or schedule for review. Not a vague recommendation. Finopscraft flags the waste and tells you what to do. Your team executes. We never auto-delete, auto-resize, or modify your cloud infrastructure.
Monthly waste identified: $465
Insights where your team already works
A forecast that lives in a dashboard nobody opens is not useful. Finopscraft pushes cost signals into the tools your team uses every day.
- Weekly forecast digest to email or Slack channel
- Real-time anomaly alerts when spend deviates from forecast by your configured threshold
- Jira ticket auto-creation for flagged waste, routed to the responsible team
- Webhook with JSON payload for custom integrations (PagerDuty, Datadog, internal tooling)
Configure per-alert thresholds to avoid alert fatigue. You decide what dollar delta or percentage deviation fires a notification.
"event": "anomaly.detected",
"timestamp": "2026-06-03T09:14:00Z",
"provider": "aws",
"service": "AmazonEC2",
"region": "us-east-1",
"current_daily_usd": 847.40,
"forecast_daily_usd": 610.00,
"deviation_pct": 38.9,
"severity": "high",
"suggested_action": "Review EC2 fleet scale event",
"alert_id": "alrt_a7f3c9"Ready to see your first forecast?
Connect your first cloud account in under 3 minutes. No agent to install.
Start Free Trial