Questions people actually ask

Not through self-serve signup, no. CoverOps is pre-launch and we are working with a small group of design partners running real production workloads on AWS or Azure.

If that sounds like you, get in touch and we will walk through what the product does today and what it does not do yet. We would rather turn you away now than sell you something that is not ready.

No. CoverOps is a control plane, not a host. Every resource it creates lives in your own AWS or Azure account, under your billing, in the regions you choose.

We operate through a role you create in your account, with a permission boundary you control. There is nothing to migrate onto and nothing of yours sitting on our infrastructure.

You keep everything. Every change CoverOps made landed as Terraform in your own repository. Revoke the role and your infrastructure carries on serving traffic exactly as before. The only thing you lose is the monitoring that finds new waste.

There is no export process because there is nothing to export: we never held it. That is a deliberate design constraint, and it is the main thing we would want you to verify before trusting us.

No, and we will not imply otherwise. CoverOps is not itself certified.

What it does is help you produce the evidence your own audit needs: enforced policy on every apply, complete change history with approvals, and exportable artefacts covering who changed what and when. Those are inputs to your audit, not a substitute for it.

It is not a replacement for engineering judgement. What it replaces is the work nobody has time for: continuously checking whether 81 cloud resources, six training runs and four agents are each earning what they cost.

For genuinely novel architecture decisions, a good engineer who understands your business will beat us. CoverOps applies known-good patterns well; it does not have opinions about your particular domain. Teams that get the most from it usually have engineering judgement in-house and want to stop spending it on plumbing.

Alongside them, not instead of them. CoverOps imports what you already run rather than demanding a rebuild, and if your existing cost tool covers cloud reporting and chargeback well, keep it. The mature FinOps platforms are genuinely better at that than we are.

What none of them cover is GPU utilisation and agent token spend, which is where we start. Onboarding never begins with destroying a working environment.

Always. Every change arrives as a pull request against your infrastructure repository, with the plan output and policy results attached.

You review it the way you review any other code. Nothing applies to production because CoverOps decided it should. The exception is the rollback path, which you arm deliberately and which only ever reverts to a state you previously ran.

Cloud accounts connect on AWS, Azure, GCP, DigitalOcean and OCI. Resource depth and cost modelling are strongest on AWS and Azure today; GCP parity is the next major piece of work.

We would rather run two clouds properly than five badly, and we will tell you honestly if your estate sits mostly where we are still shallow.

Pricing is not published while we are in early access, because we are still learning what the right shape is from design partners.

What we can tell you is the intent: a subscription, not a percentage of your cloud spend and not a share of what we recover. Both of those give us a reason to want your bill to stay large, which is the wrong incentive for a product whose whole job is shrinking it.

A small engineering team that has spent enough years on call to be unsentimental about how infrastructure work actually goes.

We are not a large organisation and we are not pretending to be. If you contact us, you will be talking to someone who writes the product.