How it works

From first login to live savings in under 15 minutes

FinOps Co-Pilot ships with a guided onboarding checklist. You connect a cloud account with a provider-native trust model, verify the data you are about to rely on, then move through waste detection, pull request cost analysis, policy guardrails, and ChatOps approvals — in that order.

The shape of it

What connects to what

Cloud accounts connect through provider-native trust models. Everything downstream — findings, forecasts, policies, approvals — runs off that one normalized dataset.

How FinOps Co-Pilot turns cloud billing data into governed actionCloud accounts connect through provider-native trust models into an ingestion and normalization layer, then into the FinOps Co-Pilot engine, which produces pull request cost checks, waste findings, forecasts, and approval-gated remediation.SOURCESINGESTENGINEOUTCOMESAWSAssumeRoleAzureService PrincipalGCPSA JWTOCINative authKubernetesHelm agentIngestionnormalize · tagreconcile · verifyFinOps Co-PilotDetect wasteForecast spendEnforce policyApex AI agentPR cost checksbefore mergeWaste findings18+ detectorsForecasts90-day horizonChatOps approvalsSlack / TeamsEvery write action is approval-gated and written to an immutable audit log
Data flows left to right: provider credentials, normalization and reconciliation, then the engine that produces every downstream signal.
Guided onboarding

Eight steps, each with a clear outcome

Every step in the checklist produces something you can verify before moving on — no long implementation project, and no waiting on a vendor services team.

  1. Step 012 min

    Create a workspace

    Start in your FinOps Co-Pilot instance and create the workspace your team will use for cloud cost operations.

    • Organization → Workspaces
    • Assign an owner and operating team
    • 2 minutes to first login
  2. Step 023 min

    Connect a cloud account

    Add AWS, Azure, or GCP using the provider-native trust model. AWS uses IAM AssumeRole; Azure and GCP credentials are stored encrypted.

    • AWS IAM AssumeRole
    • Azure Service Principal
    • GCP Service Account JWT
  3. Step 031 min

    Verify trust posture

    Validate connectivity, freshness, and trust so you know exactly which accounts are healthy before relying on cost data.

    • Verification status
    • Freshness posture
    • Detector coverage baseline
  4. Step 042 min

    Load data or ingest live

    Explore the platform immediately with demo data, or begin ingesting real spend and usage data from your connected providers.

    • Guided onboarding checklist
    • Demo or live mode
    • 2 minutes to first data
  5. Step 052 min

    Analyze a pull request

    See PR Cost Diff in action by reviewing Terraform changes before merge, with cost delta, policy, and budget signals in one place.

    • Monthly cost delta
    • Risk score
    • Policy result
  6. Step 062 min

    Run a waste scan

    Launch a live scan to surface idle infrastructure, unattached storage, rightsizing opportunities, and high-confidence savings.

    • AWS, Azure, GCP, OCI, Kubernetes
    • Composite priority scoring
    • ~2 minutes to first findings
  7. Step 073 min

    Set policies & budgets

    Establish advisory and blocking guardrails for tags, risky resource types, budget limits, and environment-specific cost rules.

    • Soft and hard budget limits
    • Tag enforcement
    • Blocking policies before merge
  8. Step 083 min

    Turn on ChatOps approvals

    Route approvals into Slack or Teams so engineering and finance reviewers can act on savings requests without leaving their workflow.

    • /finops pending
    • Approve or reject inline
    • 3 minutes to connect
Step 5, in detail

What a cost review looks like on a pull request

The highest-leverage step for most teams: catching an expensive resource while it is still a diff, not a running instance.

Pull request cost analysis posted as a GitHub commentA Terraform change adds a db.r5.2xlarge database, and FinOps Co-Pilot comments on the pull request with a plus 1,380 dollars per month delta, a policy warning, and a cheaper recommendation.FinOps Co-Pilotcommented on #482 · infra/rds.tf+resource "aws_db_instance" "main" { instance_class = "db.r5.2xlarge"~multi_az = true · storage = 500 GiB gp3MONTHLY DELTA+$1,380RISK SCOREHighPOLICYBudget warningRecommended: db.r5.xlargeSame availability profile at current utilization — saves about $620/monthEvaluated before merge · advisory policy · no infrastructure created yet
FinOps Co-Pilot comments directly on the pull request with the monthly delta, a risk score, the policy outcome, and a cheaper alternative — before anything is created.
Connection model

How each cloud connects

Every provider uses its own native trust model. Nothing depends on long-lived credentials pasted into a form.

AWS

IAM AssumeRole with an external ID. No static access keys are ever stored, and the role can be scoped to read-only until you enable remediation.

  • Cost Explorer and CloudWatch ingestion
  • Read-only until remediation is enabled
  • Per-account trust verification

Azure

A Service Principal scoped to the subscriptions you choose, with credentials encrypted at rest and never exposed in logs or UI responses.

  • Resource Graph and Azure Monitor
  • Subscription-scoped permissions
  • Encrypted credential storage

Google Cloud

A Service Account JWT scoped to billing and asset inventory, feeding spend, usage, and Recommender signals into one workspace.

  • Billing export and Asset Inventory
  • Recommender API signals
  • Project-level attribution

Kubernetes

A lightweight Helm-deployed agent that maps cluster spend to namespaces, deployments, and teams without leaving the cluster.

  • Namespace and workload attribution
  • Idle pod and orphaned PV detection
  • Works alongside Kubecost or OpenCost
FinOps lifecycle

One platform across the full cost lifecycle

Once you are set up, the same platform covers plan, observe, control, optimize, and report — replacing three to four disconnected tools.

Plan

Set budgets, policies, and governance rules before anything ships.

Budget guardrails · tag requirements · cost policies

Observe

See all cloud spend in real time across every account and team.

Multi-cloud dashboard · anomaly detection · attribution

Control

Enforce policies at PR time, deploy time, and runtime.

Advisory · blocking · tag enforcement · budget limits

Optimize

Detect waste, right-size resources, and negotiate commitments.

18+ detectors · AI remediation · Savings Plans

Report

AI-generated insights, forecasts, and executive summaries.

Prophet forecasting · team scorecards · variance analysis

Safety model

Nothing changes without a human saying yes

Detection is automated. Action is not. Every remediation passes a policy check and a human approval before it touches a resource.

Approval-gated remediation pipelineA finding moves through detection, scoring, policy evaluation, human approval in Slack or Teams, and finally execution — with every step written to an immutable audit log.Detect18+ detectors1Scoresavings · risk2Policyadvisory / block3ApproveSlack or Teams4Executeprotected tags5HUMAN GATEImmutable audit log — actor, timestamp, and payload recorded at every step
Approvals land in Slack or Teams with role verification, and every step is written to an immutable audit log.
Implementation notes

What a FinOps rollout actually involves

Most cloud cost management platforms treat onboarding as a services engagement: a scoping call, a credentials workshop, and a data warehouse build before anyone sees a number. FinOps Co-Pilot is designed for self-serve rollout, so a platform engineer can connect an AWS, Azure, GCP, OCI, or Kubernetes environment and reach live findings the same afternoon.

Start read-only, then widen scope

The recommended path is to connect one account in read-only mode, confirm the ingested spend reconciles against your provider invoice, and only then enable remediation. Because every write action is approval-gated and protected-tag aware, widening scope is a deliberate decision rather than an implicit side effect of connecting an account.

Shift cost checks left

The highest-leverage step for most engineering organizations is pull request cost analysis. Running Terraform cost estimates before merge turns cloud cost from a monthly finance conversation into a routine code review signal, which is where expensive infrastructure decisions are cheapest to reverse.

Give approvals a home

Cost findings only become savings when someone acts on them. Routing approvals into Slack or Teams — with role verification and an immutable audit trail — is what closes the loop between detection and realized savings.

Set it up this afternoon

Free trial, no credit card, and a guided checklist that gets you to live findings in about 15 minutes.