Product

Two ways to protect sensitive data.

A Chrome extension for your team. An API for your product. Both powered by the same detection engine.

Chrome Extension

Install once. Forget about it.

PII Shield runs silently in the background, intercepting sensitive data before it reaches any AI platform.

Automatic interception of all AI prompts
Real-time alerts when personal information is detected
One-click redaction to safely send your prompt
No code, no configuration — just install and go

Works with

ChatGPTChatGPTClaudeClaudeGeminiGeminiGrokGrok
Install extension
API

Build detection into your product.

Integrate personal information detection directly into your applications. Send text, get back a structured response with detected entities and a block/allow decision.

RESTful API with JSON responses
Typical detection response in one to five seconds
Canadian data residency, storage and inference
80+ entity types with combination analysis
# Send text, get structured detection results
curl -X POST https://vervian-api.fly.dev/pii/v1/detect \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"text": "John Smith, SSN 123-45-6789"}'
# Response (abridged)
{
  "should_block": true,
  "severity": "high",
  "entities": [
    { "entity_id": "E002", "type": "FIRST_NAME", "text": "John", "span": [0, 4] },
    { "entity_id": "E003", "type": "LAST_NAME", "text": "Smith", "span": [5, 10] },
    { "entity_id": "E055", "type": "SSN", "text": "123-45-6789", "span": [16, 27] }
  ],
  "triggered_combos": [
    { "comboId": "C021", "name": "First Name + Last Name" }
  ]
}

API keys are provisioned per organization during pilot onboarding. Full quickstart, error envelope, and the live OpenAPI contract on the API page.

Request API access
How it works

The detection pipeline.

Every query passes through three stages before reaching any AI service.

Stage 1

Pre-Detection

The system loads your organization's context before scanning begins.

Your organization’s settings and thresholds
80+ entity types we scan for
Your custom guardrail terms
Stage 2

Detection

Multi-layer scanning identifies all sensitive information in your text.

Regex patterns for structured data (SSN, phone, email)
AI-powered NER for contextual entities
Guardrail matching for custom terms
Stage 3

Policy Engine

Determines whether the detected entities cross your blocking threshold.

Weights every entity by severity and category
Combination analysis — the key differentiator
Final decision: allow, warn, or block
Combination Analysis

Why combinations matter.

A first name alone is rarely a privacy violation. But combine it with other data, and the risk escalates.

First Namealone
Low
First Name+Last Name
High
Full Name+Health Info
Critical

Our policy engine evaluates entity combinations, not just individual entities, to determine the actual re-identification risk.

Coverage

What we detect.

60+ entity types across 8 categories of personal information.

Personal Identifiers
Names, DOB, SSN, SIN
Contact Information
Email, Phone, Address
Financial Data
Credit Cards, Bank Accounts, Income
Health Records
Diagnoses, Medications, Provider Info
Employment Data
Salary, Performance, Job History
Legal Identifiers
Passport, Driver License, Visa
Digital Identifiers
IP Address, MAC Address, Device IDs
Biometric Data
Fingerprints, Face Data, Voice Prints
Compliance

Built around the data privacy laws protect.

Detection covers the categories of personal information these laws regulate, from health records to government identifiers to the special categories in GDPR Article 9.

HIPAA
Health Insurance Portability
United States
GDPR
General Data Protection Regulation
European Union
PIPEDA
Personal Information Protection
Canada
PHIPA
Personal Health Information Protection
Ontario, Canada
FERPA
Family Educational Rights and Privacy
United States
CCPA
California Consumer Privacy Act
California, US

Ready to get started?

Start with the browser extension or get access to the API. Setup takes under 10 minutes.