Redact a PDF with AI
AI finds the sensitive text you review every box. Handles handwriting, scans, weird fonts and very long documents, and supports any custom policy. The result document download is flattened - masked pixels with no text underneath. Free, no signup.
The download is flattened
Each redaction is rendered out of the page image and the text layer is dropped - nothing to copy, search, or recover under a box.
Pixels masked
The redacted regions are rendered out of the page image itself.
Text layer removed
No copy-paste, no find-in-page, no hidden layer for the content under a box.
Reads context, not just patterns
A name in a sentence, an identifier in a margin note - understood from context, not matched by regex.
Enterprise: private deployment
Redaction that stays inside your environment
Redaction is the workload where "send it to a SaaS" is often the wrong answer by policy, not by preference. The documents being redacted are, by definition, the most sensitive ones an organization holds - patient charts, claim files, investigative records, personnel matters. For many hospital systems, insurers, and public agencies, the data-handling rule is simple: it does not enter DocuPipe's cloud.
DocuPipe was built for that rule. For enterprise commitments, the platform deploys into your own cloud account: the API, the OCR engine, the extraction and redaction models, the workers, and the storage all run inside infrastructure you control. The documents, extracted text, and redaction boxes stay there instead of DocuPipe's cloud. Installation is a single command, and a green install includes a live end-to-end test.
Your cloud, your keys
Runs in your cloud account or VPC with your access controls, logging, and retention rules.
Full containment
API, OCR, models, workers, and storage - the whole pipeline, not a proxy back to DocuPipe's cloud.
Same product, same reviews
Your team gets the same workflows and review screens the SaaS runs - inside your environment.
Teams without an in-environment requirement run the same redaction on DocuPipe's cloud, which carries the posture this page runs on: SOC 2 and ISO 27001 certified, HIPAA compliant, encrypted in transit and at rest, documents never used for training, and BAAs executed on paid plans.
The same presets, tuned per profession
Pick the categories that match your job, add your own rules in plain language.
HIM / release of information
Producing records with other patients, family members, or restricted information in the margins - names, MRNs, DOBs, and member IDs are the daily categories.
FOIA and public records
Personal-privacy withholdings on responsive records: requester-adjacent names, home addresses, personal phone numbers, and identifiers - before anything hits the reading room.
DSAR and privacy teams
Producing one person's data means removing everyone else's. Third-party names and contact details are the default categories for a subject-access response.
Fictional samples, free to reuse: Medical record (PDF) · Incident report (PDF)
From the blog
A black rectangle is not a redaction
Why cosmetic redactions fail, and how review works here.
Read the postQuestions records and privacy people ask first
Really removed. The downloaded file is a flattened PDF - the redacted pixels are masked and the underlying text layer is dropped, so there is no live text under the boxes to copy or search.
Yes. Automated detection is an assist and can miss context a human would catch, so every proposed box appears on the real page before the file can be downloaded.
Use it as the first pass, not the sign-off. No automated tool can guarantee a legally compliant release - your reviewer makes that call. The infrastructure side is covered: SOC 2 and ISO 27001 certified, HIPAA compliant, encrypted in transit and at rest, and documents are never used to train models.
Adjust the categories or add plain-language instructions ("redact all vehicle plate numbers") and re-run - detection follows your guidelines. Whatever it proposes, the review screen is where you verify before release.
The free tool reads the first 10 pages of a document per run - redaction runs a heavier AI pass per page than plain extraction, so its cap is tighter than our other free tools. A free DocuPipe account raises the limits substantially.
Yes. For enterprise commitments, DocuPipe deploys the platform - API, OCR engine, extraction and redaction models, workers, and storage - inside your own cloud account, so documents stay in your environment instead of DocuPipe's cloud. See the private deployment section on this page.
Yes - DocuPipe workflows run redaction on every document that arrives, by API or upload, with your presets and one review queue.
Run it at scale
Redacting one file is a chore. Redacting everything your team releases is a workflow.
DocuPipe applies your redaction rules to every inbound document and gives reviewers one place to verify and release.
Free tier included. Takes about a minute to set up.
SOC 2 certified · ISO 27001 · HIPAA compliant · Encrypted in transit and at rest · Never used to train models