Built-in guardrails
Screen every prompt and response
Guardrails inspect traffic on the way in and the way out. Define them as plain topics, back them with a model asked to judge the content, or call an external service — and decide whether to block, redact or simply record.
Included
Why it matters
Stop unwanted content before it reaches a provider
Catch sensitive material in responses before it reaches users
Every guardrail leaves an evidence trail
What's included
Guardrails as topics
Describe what to watch for in plain language rather than hand-writing brittle rules.
A model as the judge
Use an LLM to evaluate traffic against your topics when simple matching is not enough.
External detectors
Plug in services somebody else runs when you need a specialist detector in the loop.
Block, redact or record
Choose the action per guardrail, and keep a record of what each one caught and did.
Ready to see it in action?
Talk to our team and we'll show you how Xcellerate AIG fits your stack.
Get started for free