De-ID.org is a browser-based HIPAA PHI de-identification engine built for healthcare AI teams, researchers, and privacy-first developers.

Most AI tools require you to upload sensitive medical documents to their servers. That creates compliance risk, audit complexity, and trust issues.

De-ID solves this differently.

It runs ONNX-based NER models directly inside the browser using ONNX Runtime Web. No PHI leaves the device. No server storage. No retention.

Upload a clinical PDF → extract structured text → detect all 18 HIPAA identifiers → download anonymized output as Markdown, JSON, or TXT.

Key capabilities:

• Client-side ML inference (true browser execution)
• PDF structure preservation (headers, tables, formatting)
• Entity confidence scores
• Rule-based cascade for SSN, MRN, regex patterns
• Progressive Web App deployment
• Zero infrastructure cost for privacy-first mode

For enterprise use cases, a cloud version powered by FastAPI + ONNX Runtime enables batch processing, audit logs, and API integrations while enforcing strict no-retention policies.

De-ID is built for:

• HealthTech startups building LLM features
• FHIR and EMR developers
• Clinical NLP researchers
• Compliance-conscious AI builders

If you're building healthcare AI and need PHI stripped before sending data to LLMs, this is the missing layer in your stack.