Protegrity AI Developer Edition
Protect Sensitive
Data from AI Workflows
Downloadable SDKs for discovering and protecting sensitive data in prompts, logs, and unstructured text.
Build and Validate
Privacy-First AI Pipelines
Composable tools for discovering, protecting, and governing sensitive data in AI-driven applications.
Data Discovery
Detect sensitive entities in raw, unstructured data.
Find & Protect
Tokenize or mask sensitive values inline.
Semantic Guardrails
Score prompt and output risk in real-time.
Synthetic Data
Generate safe datasets from real schemas.
DATA DISCOVERY
Detect PII in
Unstructured Text
Identify PII in prompts, logs, and messages with precise boundaries – even without schemas.
Input
Hi, I’m Dan Johnson and I need to update my account information.
My ssn is 123-45-6789, phone number is (555) 234-5678, and I live in LA.
Please send the confirmation to [email protected].Protegrity SDK
import protegrity_developer_python as protegrity
# Discover all PII entities (returns dict by type)
result = protegrity.discover(text)
# Extract just the text values for display
entities = {
entity_type: [d["text"] for d in detections]
for entity_type, detections in result.items()
}Output
{
"EMAIL_ADDRESS": ["[email protected]"],
"PERSON": ["Dan Johnson"],
"LOCATION": ["LA"],
"PHONE_NUMBER": ["(555) 234-5678"],
"SOCIAL_SECURITY_ID": ["123-45-6789"]
}
FIND & PROTECT
Tokenize and Mask
Sensitive Data Inline
Tokenize or mask detected entities while preserving structure and readability.
Input
Hi, I’m Dan Johnson and I need to update my account information.
My ssn is 123-45-6789, phone number is (555) 234-5678, and I live in LA.
Please send the confirmation to [email protected], thanks!Protegrity SDK
import protegrity_developer_python as protegrity
# Configure SDK with data discovery endpoint
protegrity.configure(endpoint_url="classify_endpoint")
# Find and protect PII in one call
protected_text = protegrity.find_and_protect(text)Output
Hi, I’m [PERSON]ybe B1elUnm[/PERSON] and I need to update my account information.
My ssn is[SSN]142-42-0001[/SSN], phone number is [PHONE](857) 142-4221[/PHONE], and I live in [LOCATION]iK[/LOCATION]. Please send the confirmation to [EMAIL][email protected][/EMAIL] thanks!
SEMANTIC GUARDRAILS
Score prompt and
output risk in real time
Evaluate messages in real-time and return structured risk signals your app can act on.
User message
Hi, I’m Dan Johnson and I need to update my account information.
My ssn is 123-45-6789, phone number is (555) 234-5678, and I live in LA.
Please send the confirmation to [email protected], thanks!Semantic Guardrail (SDK call)
import requests
# Scan for risks before sending to LLM
res = requests.post(
"http://localhost:8581/.../scan",
json={"messages": [{"content": user_input}]}
)
# Build assessment from response
assessment = {
"risk_score": res.json()["batch"]["score"],
"risk_level": "HIGH" if score > 0.7 else "MEDIUM",
"issues": [msg["processors"][0]["explanation"]],
"action": "BLOCK" if score > 0.7 else "ALLOW"
}Risk assessment
{
"risk_score": 0.5102,
"risk_level": "MEDIUM",
"issues": [
"OFFTOPIC"
],
"action": "ALLOW"
}
SYNTHETIC DATA GENERATION
Build safe datasets for
AI and analytics
Create statistically similar datasets from real schemas without exposing real records.
Real data
Name,Age,City,Income
Jennifer Martinez,34,San Francisco,$95000
John Smith,45,New York,$120000
Sarah Johnson,28,Austin,$75000Synthetic generator (API call)
import requests
# Generate synthetic data from real data
response = requests.post(
"http://localhost:8095/.../generate",
json={"source_data": real_data, "num_records": 1000}
)
synthetic_data = response.json()["synthetic_data"]Synthetic data
Name,Age,City,Income
Alex Chen,35,San Francisco,$93500
Michael Brown,44,New York,$118000
Emma Davis,29,Austin,$76200Docs center
Get Started in Minutes
Run locally, explore examples, and integrate with your stack.
Introduction &
Architecture
Understand how the Developer Edition fits together — components, data flow, data discovery, and data security.
Install &
Configure
Step-by-step instructions to pull the Docker images, install python modules, and get started with sample applications.
Run the
Sample app
Follow the walkthrough sample apps to see enterprise-grade data privacy in action.