LLM Security Layer — Now in Beta

Stop sensitive data
before it leaves
your perimeter.

PromptGate intercepts every LLM request — scanning prompts, responses, and file attachments for GDPR, HIPAA, PCI-DSS, and PII violations. Redact, block, or alert in real time.

See How It Works →
99%
Detection Accuracy
<2ms
Avg Latency Added
40+
Regulation Frameworks
15+
File Types Scanned
// How It Works

Proxy. Scan. Redact. Ship.

PromptGate sits inline between your app and any LLM provider — zero code changes required.

01 / INTERCEPT
Request Captured

Route your LLM traffic through our proxy endpoint. Supports OpenAI, Anthropic, Google, Azure, and any OpenAI-compatible API.

02 / ANALYZE
Deep Content Scan

Prompt text, system messages, and attached files are inspected against 40+ regulatory patterns including GDPR, HIPAA, PCI-DSS, and custom rules.

03 / ENFORCE
Block or Redact

Configure policy actions per finding type: hard block, auto-redact sensitive spans, mask file content, or emit a compliance alert to your SIEM.

04 / FORWARD
Clean Request Sent

The sanitized request is forwarded to the LLM. Your application receives the response normally, with an audit trail for every scan.

// Redaction Engine

Sensitive data, surgically removed.

PromptGate identifies and redacts PII, PHI, and financial data — in both free text and structured files — before a single token reaches the model.

Sample prompt — after PromptGate redaction
Patient PII, SSN HIPAA,
admitted 14 March 2024. Primary contact email:
GDPR. Phone: PII.

Payment on file: PCI-DSS.
Request originated from IP.
// Capabilities

Enterprise-grade guardrails,
developer-friendly setup.

Everything you need to deploy safe AI workflows — from single startups to regulated enterprises.

Real-time Prompt Scanning

Every user message and system prompt is scanned before reaching the LLM. Regex, NER, and ML-based detectors catch what rules alone miss.

File Attachment Inspection

PDFs, DOCX, XLSX, images, and CSVs are parsed and deep-scanned. Sensitive content is redacted in the extracted text before ingestion.

Response Monitoring

LLM outputs are scanned on the way back. If the model inadvertently leaks or reconstructs sensitive data, PromptGate catches it before your app sees it.

Policy as Code

Define rules in YAML or via the dashboard. Set actions per violation type: redact, block, warn, or log. Version-control your policies alongside your codebase.

Full Audit Trails

Every scan result is logged with timestamps, user context, and matched patterns. Export to your SIEM or pull via the audit API for compliance reporting.

Custom Entity Detection

Train PromptGate on your own data vocabulary. Proprietary contract IDs, internal project codes, trade secrets — define what's sensitive for your business.

// Supported Frameworks

Built for regulated industries.

PromptGate ships with pre-built detection patterns for the frameworks that matter most.

HIPAA / HITECH
PHI · PII · Medical Records
EU
GDPR
Personal Data · Right to Erasure
$
PCI-DSS v4
Card Data · CVV · Account Numbers
CA
CCPA / CPRA
Consumer Privacy · Opt-out Data
SEC
SOX
Financial Records · Audit Data
ISO
ISO 27001
Information Security Controls
US
FERPA
Student Education Records
+
Custom Policies
Define Your Own Rules
// Pricing

Simple, usage-based pricing.

Start free. Scale as your LLM usage grows. No per-seat nonsense.

Starter
$0/mo

For developers exploring safe LLM integrations.

  • 500K tokens scanned / month
  • Core HIPAA + GDPR detectors
  • Prompt + response scanning
  • API proxy access
  • 7-day audit log retention
Enterprise
Custom

For regulated industries with strict compliance requirements.

  • Unlimited token scanning
  • On-premise deployment
  • VPC / private cloud
  • SOC 2 Type II reports
  • Dedicated SLA
  • Custom retention + DLP
  • SSO / SAML / SCIM

Your LLM pipeline
deserves a gatekeeper.

Start scanning in minutes. No infrastructure changes. Just point your API calls through PromptGate.

How It Works →