Sensitive data neverreaches the model.
Feldr is a browser extension that intercepts every prompt to ChatGPT, Claude, Gemini, Copilot, and agentic AI tools — detecting and redacting PII, PHI, and secrets before they ever leave your browser.
Protecting prompts to
The problem
Your employees are the new perimeter.
Every day, people paste customer records, credentials, and financial data into ChatGPT, Claude, Gemini, and Copilot to get work done faster. None of it is malicious — and almost none of it is noticed until the data is already gone.
Customer records & health data
SSNs, medical record numbers, and customer identifiers pasted into a prompt to draft an email or summarize a case.
Secrets & API keys
AWS keys, Stripe secrets, and OAuth tokens dropped into an AI coding assistant to debug a failing request.
Financial & account data
Card numbers and account balances shared with a chatbot while reconciling an invoice or a client statement.
Source code & internal prompts
Proprietary algorithms and internal system prompts copied into a public model to get a faster answer.
Wherever your team already works
How it works
Four quiet steps — all before the AI sees a thing.
It all happens right inside your browser, in a fraction of a second. You won't notice a thing — but your private information never leaves your side.
It reads along as you type
Feldr is a small browser add-on. When you type into ChatGPT, Claude, Gemini, Copilot — any AI tool — it quietly follows along in the background.
It spots the private bits
In a split second — right on your own computer — it recognises sensitive things like names, card numbers, passwords, and health details. Nothing is sent away to check.
It hides them before you send
The instant it finds something private, Feldr blanks it out or swaps it for ▓▓▓ — before your message ever reaches the AI. You keep working; the real data stays with you.
Your team stays in control
Every save is noted in a simple dashboard, so admins can see what was protected and set the rules — without ever seeing the private data themselves.
Security & architecture
A privacy boundary enforced on every keystroke.
Every payload is detected and neutralized in the browser, then masked at the boundary — only the redacted, policy-approved version ever crosses into a prompt, a request, or a log.
Detection runs in your browser
Sensitive payloads are scanned locally by the extension and never sent to Feldr's own cloud for scanning.
Logs without the data
Every redaction is recorded as a policy-evaluated event for audit — the raw SSN, key, or card number is never stored.
Built for every AI surface
One extension install covers ChatGPT, Claude, Gemini, Copilot, and any agentic tool you open in the browser.
Detection catalog & coverage
Every sensitive pattern, cataloged and current.
48 detectors across four categories, deployed wherever your team types a prompt — and every match logged, live, in one console.
Personal data
Health records
Secrets & keys
Financial data
Browser extension
Chrome, Edge, and Firefox — intercepts every prompt typed into a web AI tool.
API & agentic surfaces
Wraps outbound calls from MCP agents and internal tools before they reach a model.
Compliance & trust
Audited, certified, and built to hold the line.
Feldr's controls are independently verified against the frameworks your security team already audits against.
SOC 2 Type II
Independently audited controls for security and availability.
HIPAA
Built to handle protected health information end to end.
GDPR
Data minimization and in-browser processing by design.
PCI-DSS
Card data is redacted before it's ever transmitted.
Get started
Stop sensitive data before it reaches the model.
Install Feldr in under a minute. Detection and redaction run in your browser from the first prompt — no setup, no plaintext leaving your machine.
Free for individuals · No credit card required · SOC 2 Type II audited