How it works, in three steps.
Know what you run, and where you run it.
The platform scans your existing accounts and reports every resource it finds: what it is, the cloud and the region it sits in, and the workload it belongs to. Nobody has to keep a list up to date.
See if it can run somewhere else.
For every workload, the list of providers it could move to, and whether it is ready to run there. Where it is not, you see exactly which services have no equivalent and need to be replaced.
Where the system actually is right now, read from your account rather than declared.
Any provider the same blueprint can deploy to today, as it is.
The services with no equivalent on that provider, named one by one.
Run it somewhere else, without rewriting anything.
In the animation you see an example of the full process: a system running on a hyperscaler is read, turned into a reusable blueprint, and deployed again on a sovereign provider, with the same setup you had. It works the same way between any supported environment, in any direction.
Four things a regulator, or a board, will ask.
Where does this data sit today?
Every running workload carries its provider and region, and the inventory comes from reading your accounts instead of asking teams to keep a list.
Could it end up somewhere else by mistake?
No. The permitted targets are declared per system and checked at deploy, so a request outside them cannot be expressed, by a person or an agent.
Who reached it, and what changed?
Every action is attributable and reversible, whether a person or an agent performed it, and the evidence is collected as it happens rather than assembled later.
What if we have to leave this provider?
You redeploy the same architecture on another provider, and the exit plan becomes a deployment choice. If you want to leave Fractal Cloud itself, everything you built exports to Terraform.