The problem Hazy solves is to give an organisation access to safe, synthetic data which is sufficiently close to the original to be equivalent for most use cases.
Rather than moving data out of the secure environment, Hazy's synthetic backbone moves Generator Models from the secure environment to users and systems within the wider organisation via the Hub.
Hazy's Generator Models are safe because they don't contain any source data. They are comprised of information about distributions and correlations within the data, not the data themselves.
These Generator Models are configured to provide certain privacy guarantees on the synthetic data they generate via e.g. differential privacy, so the organisation can choose its risk profile when assigning models to use cases via the Hub.
The Generator Models themselves are just ZIP files to facilitate automated introspection via enterprise auditing tools.