Clear authority
Approved rules run automatically. Agent proposals wait for review. Rule changes, external messages, and overrides need expert approval.
Trust
The controls are part of the record: who may approve, what ran, and which file a figure came from.
Approved rules run automatically. Agent proposals wait for review. Rule changes, external messages, and overrides need expert approval.
Trace any figure through each transformation to the file it came from, and to the approval that let it through.
Approved processes run without a model in the loop. They keep running when AI is not in use.
Any model you approve, your cloud or ours, an isolated cluster per client. Credentials stay in your vault.
AI governance
Your data is not used to train foundation models. You choose which models and gateways are available. A signed workflow does not need a model call to produce the same result tomorrow.
Your data
Audit records
Where it runs
A shared multi-tenant cloud can stall legal review, residency commitments, and internal security policy. Fontana runs where your reviewers will sign off: in your cloud or ours, or on your own infrastructure. Identity is SSO, MFA, and SCIM in every workspace. Agent code runs in short-lived sandboxes. Changing a permission is a policy change: approved, versioned, and recorded.

Assurance
Sensiba and Tempo examined Fontana. The SOC 2 Type 1 report, issued in 2026, is available to customers and prospects under NDA. Control narratives, our DPA, and the subprocessor list are available the same way.
Request the reportRead the control summary