Enterprise Guardrails

Protect your LLM usage with content guardrails. Detect and block prompt injections, PII, secrets, and more — available for Enterprise customers.

Guardrails configuration dashboard

Guardrails protect your organization by automatically detecting and blocking harmful content in LLM requests.

Built-in Protection

System rules detect common threats:

  • Prompt Injection — attempts to override system instructions
  • Jailbreak — attempts to bypass safety measures
  • PII Detection — personal information like emails, phone numbers, SSNs
  • Secrets — API keys, passwords, and credentials
  • Sensitive File Types — block uploads of restricted file types

Configurable Actions

For each rule, choose how to respond:

  • Block — reject the request entirely
  • Redact — remove sensitive content and continue
  • Warn — log the violation but allow the request

Custom Rules

Create organization-specific rules:

  • Blocked Terms — prevent specific words or phrases
  • Custom Regex — match patterns unique to your use case
  • Topic Restrictions — block certain topics

Security Events Dashboard

Monitor all violations with a dedicated dashboard showing:

  • Total violations and trends
  • Breakdown by action taken (blocked, redacted, warned)
  • Detailed violation logs with timestamps

Access

Guardrails are available on the Enterprise plan.

Interested? Contact us to enable Enterprise for your organization.