OpenAI

AI·AI Outputs & Agent Actions

Connect Aguardic to OpenAI to enforce runtime guardrails on GPT model outputs.

What You Can Govern

Apply policies to every aspect of this integration

GPT-4, GPT-3.5 model responses

Prompt and response content

Model parameters and settings

Token usage and costs

User context and metadata

Choose how OpenAI enforces your policies

Block

Block response, return policy violation message to user

Approval

Hold request — return review ID for human approval

Warn

Allow response but log warning for review

Monitor

Log all interactions without blocking

Pre-built policy templates you can enable for OpenAI

Setup Guide

Get OpenAI running in minutes

Add OpenAI API key in Aguardic

Route OpenAI requests through Aguardic proxy

Select policy sets for LLM governance

Set enforcement mode

Test with sample prompts

Permissions

What Aguardic needs to govern OpenAI

Proxy API requests to OpenAI

Read request and response payloads

Monitor model usage

Connect OpenAI and start enforcing policies in minutes.