Skip to main content
StarfortStarfort Docs

Control how your organization uses AI.

Starfort puts guardrails between your people, your apps, and the AI services they use — masking sensitive data, blocking disallowed topics, and recording every decision for audit.

Ask anything about Starfort…Coming Soon

Admin Console

Set up organizations, Guardians, and Guard Policies end-to-end.

Guard API

Embed guardrails in your own services with one POST request.

Desktop Agent

Protect everyday AI usage on employee PCs — transparently.

What is Starfort?

The platform, the risk model, and the core concepts in five minutes.

Glossary

Every Starfort term used in these docs, defined.
v1.4
  • Stratum — a new module that classifies files (documents) by sensitivity grade. Define your grading criteria with a Taxonomy and a Rule Package; classification runs through an L1–L4 cascade. For console setup, see Stratum policies.
  • Module Instances — a generalization of Project Guardian. A project now runs Project Guardian and Project Stratum side by side as peer instances.
  • S3 de-identification batch — an asynchronous batch API that de-identifies files in S3 buckets connected through a Storage Connection.
  • Processed-content egress — everything that leaves for an external AI service is now the processed content (processed_content). Even on PASS, what’s forwarded is the content Guardian actually inspected and rebuilt — not the original.