Your Users Are Pasting Sensitive Data Into AI. Stop Them.
Screen prompts, uploads, and RAG documents before sensitive data reaches your model. Deterministic detection in milliseconds—not probabilistic guessing.
Data You Never Collect Can't Be Breached
Users paste SSNs into support tickets. Upload spreadsheets with customer PII. Submit prompts with confidential data. Once it hits your system, you own the liability.
Support & Helpdesk
Customers paste account numbers, SSNs, and medical info into tickets constantly. Every message is a potential compliance violation.
AI & LLM Applications
Users paste confidential data into AI prompts. Your RAG pipeline ingests PII. Once it's in the model, it's retrievable.
File Uploads
Resumes with DOB and SSN. Contracts with confidential terms. Spreadsheets with customer data. All landing in your storage.
How It Works
One scanning engine. Multiple integration points.
User Submits Content
Form submission, file upload, or AI prompt—any user input triggers a scan.
Real-Time Scan
250+ classifiers check for SSNs, credit cards, PHI, credentials, and more. Milliseconds, not seconds.
Take Action
Block the submission.
InspectShield.scan(formData)
.then(result => {
if (result.hasSensitiveData) {
// Block or warn
showWarning(result.findings);
} else {
// Safe to submit
submitForm(formData);
}
});Integration Options
Same scanning engine, multiple entry points
Drop into any web form, chat UI, or file upload. Scans client-side before data leaves the browser.
One-click install. Protects Contact Form 7, Gravity Forms, WooCommerce, and any form plugin.
Server-side scanning for RAG pipelines, backend uploads, webhooks, and data ingestion workflows.
Run the scanning engine in your environment. Zero data egress. Full control.
Built for the AI Era
Every AI application needs a data firewall. Screen prompts before they hit the LLM.
Deterministic, Not Probabilistic
LLMs guess whether something is sensitive. Pattern matching knows. When compliance is on the line, "80% confident" isn't good enough.
Inspect-Shield: "This IS SSN: 123-45-6789"
Save AI Cycles (and Cost)
LLMs are slow and expensive for document inspection. Pre-screen with fast, cheap classifiers. Only send clean content to the model.
LLM inference: ~2000ms, $0.01+
RAG Pipeline Hygiene
Don't let PII into your vector database. Once it's embedded, it's retrievable. Screen documents before ingestion—not after your chatbot surfaces a customer's SSN.
Audit Trail
"What sensitive data did users attempt to send to our AI?" Deterministic scanning gives you logs and evidence. LLM-based detection gives you... confidence scores.
Who Uses Inspect-Shield
Any application that accepts user input
AI/LLM Applications
ChatGPT-style interfaces, copilots, and AI assistants. Screen prompts before they hit the model.
Support & Helpdesk
Zendesk, Intercom, Freshdesk, or custom ticketing. Stop customers from oversharing.
HR & Recruiting
Job applications, employee portals, benefits enrollment. Resumes are PII goldmines.
Healthcare Portals
Patient intake forms, document uploads, messaging. HIPAA compliance at the front door.
Financial Services
Account applications, document submission, customer communication. Every form is a risk.
SaaS Platforms
Any app with file uploads, comments, or user-generated content. Limit your liability.
250+ Classifiers, Zero Configuration
All classifiers run simultaneously. No picking and choosing. No missed detections.
Frequently Asked Questions
How is this different from using an LLM to detect sensitive data?
LLMs use probabilistic detection—they guess. Inspect-Shield uses deterministic pattern matching—it knows. When compliance is on the line, you need certainty. Plus, pattern matching is 100x faster and cheaper than running every input through an LLM.
What integration options are available?
Three options: (1) JavaScript SDK that drops into any web form or AI chat interface, (2) WordPress plugin for instant protection, (3) REST API for server-side scanning of RAG pipelines, backend uploads, or any data flow.
Can I self-host Inspect-Shield?
Yes. The scanning engine runs as a Docker container in your environment. Your data never leaves your infrastructure. We also offer a hosted API if you prefer not to manage infrastructure.
What happens when sensitive data is detected?
You choose: Block the submission entirely, warn the user and let them decide, or automatically redact the sensitive content before it reaches your system. All configurable per-classifier.
How fast is the scanning?
Milliseconds. Pattern matching is nearly instantaneous compared to LLM inference. Users won't notice any delay when submitting forms or uploading files.
What types of sensitive data can Inspect-Shield detect?
250+ classifiers covering SSNs, credit cards, bank accounts, driver's licenses, passports, medical record numbers, API keys, passwords, and more. Covers HIPAA, PCI, GDPR, CCPA, and other regulatory requirements.
Stop Collecting Data You Don't Need
The cheapest breach to remediate is the one that never happened.
Get early access to Inspect-Shield.