AI DETECTION + HUMAN REVIEW

Redact what matters.
Keep what doesn't.

DocArmor.ai finds every SSN, MRN, and name buried in your PDFs, then hands you the final call before anything gets burned into the file.

Self-hosted or cloud Full audit trail No data used for training
deposition_transcript_0417.pdf
RE: Case No. 22-CV-04471 — Witness Statement
Deponent: Sarah KesslerNAME, residing at
4417 Ashgrove Ln, Nashville, TNADDRESS
DOB: 03/14/1988DOB   SSN: •••-••-4471SSN
Testimony begins on the following page —
HOW IT WORKS

Three steps between a raw file and a clean one.

Nothing gets redacted or anonymized without a person signing off first.

01 / INGEST & DETECT

Drop in the file

PDFs, scans, exports — the AI pipeline reads every page and flags names, IDs, dates, signatures, and faces as it goes.

02 / HUMAN REVIEW

Review every flag

Each detected entity lands in the HITL console. Accept it, adjust the boundary, or add a redaction the model missed.

03 / BURN & EXPORT

Lock it in

Approved redactions are physically burned into the file, or entities are swapped for consistent aliases. No layer to peel back.

PRODUCT

Built for the review, not just the scan.

Detection is the easy part. DocArmor.ai is built around the review step that actually makes redaction defensible.

PII & PHI detection

Entity recognition tuned for names, identifiers, dates, addresses, and signatures across typed text, scans, and tables.

HITL review console

Step through flagged entities side by side with the source page. Keyboard-driven, built for reviewing hundreds of pages a day.

True redaction burn-in

Approved redactions remove the underlying content, not just paint over it. Nothing left to copy-paste or reveal.

Anonymization mode

Swap sensitive entities for consistent aliases instead of blacking them out, so the document stays readable for analysis.

Audit trail

Every accept, edit, and rejection is logged with the reviewer and timestamp — ready for a compliance file.

Batch & API

Point it at a folder or pipe documents in through the API when the review queue needs to run at volume.

INDUSTRIES

Wherever a document holds someone's data.

LEGAL

Discovery & depositions

Clear privileged and personal information from case files before production, with a record of every call made.

HEALTHCARE

Patient records

Strip identifiers from charts and referrals so they can move between systems and studies without exposing a patient.

FINANCIAL SERVICES

Loan & KYC files

Prepare account statements and applications for audit, dispute review, or a third party without the account holder's data riding along.

HR & GOVERNMENT

Personnel & FOIA

Respond to records requests with reviewer sign-off on every redaction, not a black marker and a hope.

SECURITY

Privacy-first, by construction.

DocArmor.ai is built for teams who can't send sensitive documents through just anything. Deploy it in your own environment, or run it in ours — the review workflow is identical either way.

Self-hosted or VPC deployment

Keep documents inside your own infrastructure when that's a requirement, not a preference.

Encrypted in transit and at rest

Documents and detected entities are encrypted end to end, at every stage of the pipeline.

No training on customer data

Nothing you upload is used to train models, ours or anyone else's.

Role-based access & audit logs

Control who can review, approve, or export, with a full history of who touched what.

Stop redacting by hand.

See DocArmor.ai flag, review, and burn in a real document in about ten minutes — yours, or one of ours.