What's new in KeptPDF

The Print button and Ctrl+P did nothing. Our own security rules blocked the way we handed the document to the printer, and the failure was silent, so nothing happened and nothing explained why. It was worst in the installed app, where there is no browser menu to fall back on. KeptPDF now draws every page itself and prints those, so the button works the same in a tab and in the installed app. A large file takes a moment to get ready, and a small bar shows how far along it is, with a Cancel button if you change your mind. Your file still never leaves your device.

Suggested splits already found where one long scan is really several documents. Now each one wears its type: Permit, Application, Inspection, Certificate, Survey, Property card, Recorded and six more. It reads the body of the page rather than the biggest line on it, because the biggest line is usually a letterhead. When it is not sure, it says so quietly instead of filing a guess, and you can click the type to change it. A new switch in the Done menu, "Sort into folders by type", files the output the same way: one folder per type inside the ZIP, or one heading per type in the sidebar of the single bookmarked PDF.

The same redaction engine, now a command. npm install -g keptpdf and point it at a folder. It burns the marked text out of the page, re-reads every file it wrote, and leaves a certificate beside each one. First 100 files are a trial, with no account and no card. A license is $4,800 per organization per year, unlimited files. Details at https://keptpdf.com/cli.

keptpdf review writes a spreadsheet you edit before the burn. keptpdf watch is a drop box on a machine you already trust. Auto-detect is a start, not a guarantee: a person still reads the output before it goes out.

Exemption codes are yours to choose, with nothing preselected. Stamp a run with --label "(b)(6)", or set a different code on a single box. The browser tool and the CLI both do this.

Prove a redaction actually held. Drop in a PDF and KeptPDF finds black boxes that never flattened, recovers any text still hiding underneath, and surfaces sensitive details you have not redacted yet. Free shows an on-screen risk summary with the recovered text inline; Pro exports a full Exposure Report. One click sends any finding straight into Redact to fix it, and teams can X-ray a whole folder into a single report.

Re-flow any PDF into a clean, phone-friendly reading view, with real bold and italic from the document's own fonts, tables rendered as scrollable grids, an OCR pass that quietly repairs scan errors (and lets you audit each fix), keyboard page navigation, and an in-document Find. Free, with no quota.

Built for data-subject-access requests. Tell Redact who the request is about and it keeps that person's details while removing everyone else's, with a dedicated subject panel and a matching section on the audit certificate.

The reversible tool that swaps names, dates and account numbers for consistent aliases, so you can paste a document into ChatGPT or Claude and still map the answer back, now has its own page. You can hand-draw your own blocks and name each one. Reversible on your device; nothing uploaded.

Bring in any photo or camera capture and KeptPDF turns it into a searchable PDF, ready for any tool in the suite.

A single hub at /trust for the proof behind the privacy promise, plus a downloadable Security and Privacy Overview, now linked from the top navigation on every page.

Tap a word to box it, no dragging required, though dragging still works across words and rows. Padding is adjustable, white-out fill is one toggle, and tapping a single match offers to redact every other occurrence too.

Detects handwritten signatures and handwriting, locating them rather than reading them, even on photographed pages, and now reads sensitive text out of annotations and inside images, with an honest prompt to review anything it could not read for itself.

Auto-detect now finds European day-first dates, ISO-8601 timestamps, comb-field SSNs and account numbers on fillable forms, loan and mortgage IDs, medical accession numbers, masked SSN last-fours, and names scattered across flattened forms.

A large real-world accuracy pass stopped roughly 600 bad matches. Place names, job titles, agency and product names and lab reference ranges are no longer redacted as people, and web links, toll-free numbers and standalone dates now go to review instead of being removed automatically.

A spotlight glides smoothly to each match and stays locked on the word as you zoom, uncertain matches get a dashed amber outline, and a "Review complete" toast glides you back to the start of the document when you finish.

Pro can save and reuse personal redaction-term lists, stored locally with opt-in end-to-end-encrypted sync. Teams can share encrypted lists with admin controls.

The audit certificate now tallies what was removed by origin (form field, annotation or image) and carries a clickable verify link, and redaction strips embedded-image EXIF and GPS data on the way out.

A floating page indicator and scroll-past-the-edge paging now work the same way across Redact, Sign, Annotate and Compare.

The editors fill the whole window, and Back now returns you to the tool's setup screen instead of dropping you out of the tool.

Rotate placed objects and redaction boxes to any angle, with a single rotate handle in the same place across the editors.

Done now finalizes and downloads in a single click, and your finished file downloads automatically.

Highlight in Annotate by clicking a word or dragging, just like Redact, plus a Focus differences fade slider in Compare, a tidier toolbar, multi-line watermarks, and an ink-saver mode in Compress that whitens colored backgrounds while keeping the text crisp.

The metadata report is now one card graded by severity, and a quiet cross-tool peek warns you about hidden data in a file before you share it.

Drop a long, tab-less PDF and get nested bookmarks plus a clickable contents page, built from the document's headings, all on your device. Tuned for bills, contracts and reports.

Rename a stack of scanned PDFs from the text inside each, pulling date, party and document type into a template, with OCR for scans and rename-in-place where the browser supports it. Free covers 3 files per batch.

Swap names, dates and account numbers for consistent aliases so you can paste a document into ChatGPT or Claude and still map the answer back. Reversible on your device; nothing uploaded.

Label documents Exhibit A, B, C with divider sheets, a combined binder and an exhibit index, free as a text list. Pro adds the formatted PDF and CSV index.

On-device OCR now straightens skewed pages and adapts to phone photos and low-contrast scans before reading them, cutting recognition errors.

The signature pad draws with a variable-width ink stroke and goes full-screen on phones, so a drawn signature looks like a signature.

Auto-detect now catches more real names, including middle initials and ALL-CAPS forms, while dropping more transaction-noise false positives.

Run your everyday tools across a whole stack of PDFs in one pass: redact, Bates-stamp, OCR, compress, scrub metadata, watermark or password-protect. Every file is processed on your device; Free covers 3 files per batch, Pro and Practice are unlimited.

KeptPDF finds a form's fields and drops a box on each. Click to type, check, date and sign, then flatten. Flat forms get click-to-place fields too.

Turn bank statements and invoices into a clean grid. Review it, then export to CSV (free) or Excel (Pro). Nothing is uploaded.

Auto-detect now flags postal addresses and ZIP codes, web links, and special-category details such as health and financial information, alongside expanded coverage of URLs, ages over 89, and device serial numbers for HIPAA Safe Harbor.

Start from a profile, such as Safe to share, Court (FRCP 5.2), Financial, or Contact, and KeptPDF turns on the right categories for the job. Adjust any of them and it remembers your custom set.

Type a name, case number, or any phrase and redact every occurrence across the document, combined with auto-detection in a single review.

A full litigation workflow: per-custodian prefixes, confidentiality designations, slip sheets, "skip" pages, and review-platform load files (.DAT / .OPT) with a production report.

Send one tool's output straight into the next, for example Redact → Bates, without re-uploading or re-selecting your file.

A new /hipaa page with a downloadable pack (Technical Privacy Brief, BAA template, and compliance FAQ) for teams that need the paperwork.

Auto-detect now finds ITIN, EIN, routing, IBAN, NPI, DEA, and VIN numbers, plus label-anchored passport, driver's license, MRN, and name detection.

A big accuracy pass on real-world PDFs, plus an OCR text-layer fix so trailing characters can no longer leak past a redaction box.

Every redaction now produces a certificate of what was removed. Free: a text certificate with categories and an input hash. Pro: a tamper-evident PDF with a Safe Harbor table plus a JSON sidecar.

The PDF engine, viewer, and OCR are now fully self-hosted. KeptPDF loads nothing from outside servers, so the "it all stays on your device" promise is verifiable end to end.

Straightforward Free / Pro pricing and a corrected, benefit-by-benefit comparison against other private PDF tools.

A new highlight tool with snap-to-text, alongside a more compact editor toolbar.

Save more than one signature and give each a color tint, so co-signers stay distinct.

Tap to select, pinch and trackpad zoom anchored under your cursor, and two-finger panning across the editors.

After each operation you now get a page thumbnail, a clear receipt of what happened, and per-action icons.

Your recent files stay in view and no longer scroll out of reach.

Scrollbars and form controls now follow the dark theme instead of staying stubbornly light.

A quiet prompt to save your work, shown only after a couple of uses in a day.