Getting started

Introduction

What DataStrict is, who it's for, and the problems it solves.

Overview

DataStrict is the control layer for modern AI systems. It sits between your models, your data, and your users - turning written governance into deterministic enforcement.

Every prompt is evaluated against policy. Every tool call is bound to identity, purpose, and jurisdiction. Every output is inspected before it leaves the boundary.

Who it's for

Platform, security, and AI teams at regulated enterprises - financial services, healthcare, insurance, public sector - who need to ship AI features without taking on unbounded risk.

Tip · Not a wrapper

DataStrict does not replace your model provider, your data warehouse, or your identity platform. It composes with them - adding the enforcement layer they don't ship.

Architecture

DataStrict deploys as a sidecar gateway in your infrastructure. SDKs are available for Python, TypeScript, and Go. The control layer is single-tenant by default for enterprise deployments, with BYOC available for regulated environments.

client.completion({
  model: "gpt-4o",
  messages: [...],
  policy_context: { user_id, purpose: "support" }
})

Next steps

Continue to the Quickstart to enforce your first policy in under 10 minutes.

Was this page helpful?

Suggest an edit

Back

Home

Quickstart