Guardrails is an open-source Python framework for adding guardrails to applications built on large language models. It runs input and output guards that detect and mitigate risks such as toxic language, personal data or competitor mentions, and it helps generate structured output based on a Pydantic model. Validators come from the Guardrails Hub and are installed as packages. You use it as a Python library or run it as a standalone server with a REST API. The company Guardrails AI also offers Guardrails Pro, a managed service with a private hub and observability dashboards.
The framework is licensed under Apache 2.0 and can be run entirely under your own control, keeping data in your own environment. The company Guardrails AI is a US startup based in San Francisco. According to the vendor, Guardrails Pro is deployed as a centralised service inside the customer's own VPC. For European organisations, self-hosting is the direct route to control over data and jurisdiction; the paid service still comes from a US vendor.