OpsTillerDetect
Combine billing, utilization, tags, and account context into a defensible finding.
AWS cost remediation for engineering teams
Turn AWS waste into reviewed fixes.
OpsTiller will connect cloud-cost findings to their owners, route safe approvals, and prepare narrow Terraform pull requests—so savings can move from backlog to production.
A workflow, not another list
From signal to pull request.
Explore how a future OpsTiller workflow could turn an unused load balancer signal into an evidence-backed, owner-approved change.
Remediation workflow Workflow OT-0142
Evidence
Unused load balancer detected
Thirty days with no requests or healthy targets support this finding.
Requests0
Healthy targets0
Lookback30 days
1 / 5
The missing middle
Finding waste is the easy part.
Most recommendations stop exactly where engineering work begins: proving the signal, finding the owner, changing code, managing risk, and checking the result.
Today’s backlog
- 01 Generic recommendation
- 02 Unknown service owner
- 03 Screenshot in a ticket
- 04 Savings never verified
→
With OpsTiller
- 01 Evidence and confidence
- 02 Terraform ownership match
- 03 Approval-gated draft PR
- 04 Observed savings window
How it would work
Guardrails at every step.
A deliberate path from observation to action, built to fit existing engineering controls.
Verify
Recheck evidence and match the live resource to its owner and Terraform source.
Remediate
Route an approval and prepare a narrow draft pull request—never merge or apply it.
Measure
Compare the accepted opportunity with observed billing data after deployment.
Initial catalog
Conservative by default.
RDS
Idle RDS instances
Correlate connections, CPU, environment, and ownership before proposing a reviewed change.
- Potential
- $412/mo
- Risk
- Medium
- Signal
- 14d evidence
S3
Missing lifecycle policies
Identify aging objects and draft a scoped Terraform lifecycle policy for review.
- Potential
- $186/mo
- Risk
- Low
- Signal
- 90d objects
EBS
Unattached EBS volumes
Trace detached volumes to owners, preserve recoverability, and route an approval.
- Potential
- $94/mo
- Risk
- Medium
- Signal
- 30d detached
Prompt to policy
Describe the outcome. Review every rule.
Natural language could become a validated, versioned workflow—not arbitrary production code.
“Find internal load balancers with no requests or healthy targets for 30 days. Exclude protected resources and require platform approval.”
unused-internal-alb.yaml
scope:
environments: [non-production]
excludeTags:
cost-remediation: protected
detector:
type: idle_resource
lookbackDays: 30
thresholds:
requestCountMaximum: 0
healthyTargetCount: 0
action: terraform_pull_request
approval:
required: trueSafety is the product
Designed for earned trust.
OpsTiller starts with constrained access and explicit control—not an autonomous agent with production credentials.
Read-only first
Discovery would begin with temporary credentials from a customer-deployed, read-only AWS role.
Approval is explicit
Every proposed mutation would bind an approver to a specific preview and immutable workflow version.
Terraform stays authoritative
Owned resources would change through draft pull requests. OpsTiller would never merge or apply them.
Stale findings stop
Evidence would be revalidated immediately before a proposed action proceeds.
Actions are allowlisted
Direct API actions, if introduced, would use narrow adapters with declared permissions and preconditions.
Every outcome is auditable
Findings, decisions, workflow versions, and verification status would retain a clear history.
A note from the founder
Help us build the part that actually ships.
Cloud teams rarely need one more dashboard telling them what they already suspect. They need a trustworthy way to move from opportunity to owned, reviewed work.
We’re speaking with platform and infrastructure leaders about the last optimization they tried to ship: where it stalled, who had to approve it, and what proof made the change safe.
Book a 25-minute conversation →FAQ
Questions, answered plainly.
Is OpsTiller available today?+
Not yet. OpsTiller is in product discovery. We’re inviting a small group of engineering teams to shape a read-only alpha and validate its remediation workflows.
Will OpsTiller make production changes?+
Not autonomously. Terraform-owned resources would be changed through draft pull requests, with explicit human approval and your existing CI controls.
What access would the alpha require?+
The proposed alpha would use a customer-deployed read-only AWS role and a GitHub App. Remediation permissions would be separate and optional later.
How is this different from another cost dashboard?+
The product is being designed around the work after detection: evidence, ownership, approvals, reviewable changes, and verification.
Who is it for?+
AWS-heavy SaaS teams using GitHub and Terraform, especially platform, infrastructure, DevOps, and FinOps leaders responsible for cloud outcomes.
Early access
Make cost optimization
an engineering workflow.
Join the waitlist for product updates and an invitation to the read-only alpha.