Waste, pricing, AI, tickets, Terraform and the security add-on. Each piece is below. Pick one and jump.
30 days, no card. Read-only connection.
51 deterministic rules across 3 providers. Detects unused, oversized, or misconfigured resources. All included in Pro, Business and the 30-day trial.
Sample findings
Illustrative dataEC2 with CPU < 5%
FreeEC2 instances with average CPU below 5% over the last 7 days. Candidates for rightsizing or shutdown.
Orphaned EBS snapshots
FreeSnapshots whose source volume no longer exists. They occupy space and generate cost with no utility.
Unassociated Elastic IPs
FreeElastic IPs reserved but not assigned to any resource. AWS charges for each unused IP.
Unattached EBS volumes
FreeStorage volumes disconnected from any instance. You pay per GB/month without using them.
Reserved Instances, rightsizing engine, and scenario simulator. Everything you need to pay less for the same infrastructure.
Analyzes EC2 and RDS instances with stable usage and recommends 1-year or 3-year commitments with monthly savings projection in USD.
For each oversized instance (CPU p95 < 40%), suggests the smallest viable type with 1.3x headroom.
Interactively calculates savings with 4 configurable scenarios: turn off dev at night, buy reservations, apply rightsizing, or move cold S3 data to Glacier Flexible Retrieval.
Not only what is public. Network, IAM, encryption, logs and secrets. €20 add-on per account on Pro and Business. Not in the trial.
Score
72
Public
4
Extra access
3
Network
Open
Identities
Open
Encryption
In order
Logs
Review
Secrets
In order
Backup bucket visible
List + Get for anyone
SSH open to the internet
Also idle
Account with extra admin
3 people with * in IAM
URLs, health endpoints, AWS, Azure and GCP resources and crons with no agents to install, and flows that point to the exact step that broke.
Monitors
Illustrative data
api.example.com/health
Health contract
Up · 120 ms
orders-in
SQS queue
Slow · 4 min
nightly-billing
Heartbeat
Called 3 min ago
order-to-invoice
Flow
Broken at step 2
The spend or outage alert hits the channel. A finding can open a ticket. No copy-paste.
Today
Illustrative data
Slack (webhook)
Color-coded attachments by severity, finding summary, and direct dashboard button.
Jira/Linear tickets
Automatic ticket creation when alerts fire. Configurable minimum severity. Secure API token.
Jira/Linear tickets
Automatic ticket creation when alerts fire. Configurable minimum severity. Secure API token.
From PDF/CSV export to advanced reports with chargeback, 12-month forecast, and anomaly detection with root cause.
Tagging coverage per tag and spend by region, resource type and service.
Cost attribution by 5 tag dimensions with progress bars.
Seasonality, confidence intervals, and detailed monthly table.
Root cause of spend spikes with specific resource and suggested action.
Sample report
Illustrative dataThis month
PDF export
ProCSV export
Pro3-month forecast
Pro12-month history
ProA normalized resource schema allows analyzing any provider with the same rules and AI engine. One panel for everything.
1-click CloudFormation template. Read role with ViewOnlyAccess, SecurityAudit and Cost Explorer, plus ExternalId.
Service Principal with Reader role. Multi-subscription support.
Service Account JSON with Viewer role. Multi-project.
The same scan you already run attributes the cost of your AI and ML services across AWS, Azure and GCP, broken down by resource and mapped to your tags. Nothing new to connect.
Amazon Bedrock (provisioned throughput) · SageMaker Endpoints · SageMaker Notebook Instances
Azure OpenAI Service · Azure Cognitive Services · Azure AI Search
Vertex AI Endpoints · Vertex AI Notebooks (Workbench) · Vertex AI Training Jobs
The company account connects with an admin key: spend by project, model and person, alerts, and where the plan includes it, a fix with approval. Each person's account needs no key. The cost is the plan's list price. Usage is the quota percent the team declares, not an invented amount. On Cursor, Claude Code and Codex, the agent on the computer also reports tokens by model, and on Copilot requests, without adding them to the invoice.
On the company account (OpenAI, Anthropic, Copilot, Cursor or OpenRouter) you see daily spend by project, model, seat or person, in the same scan and the same reports as your cloud.
An employee's ChatGPT, Claude, Cursor or Copilot plan does not expose spend. You register the email, the plan and the renewal day. We show the list price and the quota percent the team declares. We do not invent tokens or add them to the invoice. On Cursor and Claude Code, that agent reports tokens by model and does not add them to the invoice. Copilot reports requests by model through gh, and the GitHub token stays on the computer. Codex reports tokens only for a ChatGPT subscription, not an API key.
We detect underused caching, oversized context windows, expensive models for simple tasks, and missing spend alerts or limits.
On providers that support it, we create the spend alert or limit in one click, with a grace period and rollback. When the provider has no such API, we give you the exact manual guide to fix it yourself.
5 predefined roles, granular permissions per role, up to 50 members, and client management for agencies/MSPs.
The same public API: OAuth2, an MCP server for chat, and a Terraform provider so accounts and alerts live in code.
Open the docsCreate apps with client_id/client_secret. POST /oauth/token endpoint returns 1-hour HS256 JWTs with granular scopes.
Ask Claude Desktop, VS Code or Cursor which account overspends or how secure it is. Fourteen read-only tools, over stdio.
Provider in the AAA Sentinel registry. It does not replace the Terraform that builds your cloud: it manages AAA Sentinel.
Shutdown schedules, FinOps playbooks, posture remediation on AWS, Azure and GCP, and AI spend controls. Write is separate, with approval.
Business plan. FinOps and posture on AWS, Azure and GCP.
See every automation