> ## Documentation Index
> Fetch the complete documentation index at: https://docs.synscribe.com/llms.txt
> Use this file to discover all available pages before exploring further.
# How to automate compliance and quality checks for your content
> The compliance guardrail is an AI reviewer that checks every piece against your rules and blocks a publish that breaches one, so nothing off-limits goes live.
Synscribe has a built-in compliance guardrail: an AI reviewer that reads every landing page and blog
post against the rules you set, and blocks a publish that breaches one. You write your rules once and
it enforces them on every draft, so content stays inside your legal and regulatory lines without you
hand-checking each piece. It's off until you turn it on, and you switch it on per content type. In
the app it's the [Compliance Guardrail](https://www.synscribe.com/app/guardrail).
## When you need it
If you're in a regulated industry, or your marketing goes through legal review, you have claims you
can't make and language you're required to include. That's what the guardrail is for. Instead of a
person checking every page against a do-not-say list, you encode the list once and the guardrail
enforces it at publish, every time. Common cases:
* Payments, crypto, or financial services: no guaranteed returns, no implied regulator endorsement, no absolute-safety claims.
* Health or wellness: no unsubstantiated medical claims, required disclaimers.
* Any business with legal sign-off on marketing: the constraints your counsel already gave you, encoded as rules.
The rules are yours to write, so the same mechanism enforces content-quality standards too, not just
legal ones. Whatever you'd otherwise catch in a manual review, you can make a rule.
## Guardrail vs refiner
Two systems touch compliance and they do opposite jobs, so it's worth knowing which is which. A
**refiner rewrites** content: brand positioning, CTAs, or compliance edits, and the
[landing-page refiner](/platform/landing-page-refiners) applies schema-safe compliance edits across
many pages at once. The **guardrail doesn't edit anything**. It judges pass or fail and stops a
non-compliant piece from shipping. They pair up: the guardrail catches a problem at publish, and its
suggested fix feeds a one-click refiner edit. The full comparison is on
[Refiner vs guardrail](/platform/refiner-vs-evaluator).
## How it works, start to finish
1. [Write a policy](/platform/set-up-evaluator): your compliance context plus a few atomic rules, each stating what to flag and what not to.
2. [Calibrate it](/platform/calibrate-evaluator): run it on copy you know should pass and copy you know should fail, and tighten the rules until it gets both right.
3. Turn it on for blog posts, landing pages, or both. Nothing is gated until you do.
4. [It gates at publish](#how-it-blocks-and-where-you-see-it): a breach blocks the publish and shows you the violations, so you fix them or approve the exact version.
For the full compliance-and-quality setup, see
[how to set up Synscribe for compliance and quality control](/platform/compliant-content-regulated-industry). To
understand how the judge reaches a verdict, see
[how the guardrail decides](/platform/how-evaluator-decides).
## How it blocks, and where you see it
When the guardrail is on for a content type, it runs the moment you publish. Content that passes goes
straight through; content that fails is held, with the violations shown, so nothing off-limits ships
by accident. You can approve a specific version to let it through, and editing a piece re-checks it.
You see the result right on your content. Blog posts show a **Guardrail** status in the Content
Library, and a blocked piece opens its violations with the exact excerpt and a suggested fix.
Open a blocked piece to read the flag and fix it, or **Approve** that exact version to publish it
as-is.
Landing pages are gated the same way when you publish or export them: a failing page is held with its
violations, just like a blog post.
**Next:** [Set up the compliance guardrail](/platform/set-up-evaluator)