Éditeur PDF avec IA v1.0

An AI PDF editor is a tool that changes a PDF based on a plain-language request instead of menus and toolbars. With the DocPivot AI PDF Editor, you type something like "delete page 3, rotate pages 5-7, stamp APPROVED on page 1, then compress," check the step list the AI builds, and apply it. The AI only plans the work. It never receives your file, most edits run inside your browser, and you get 25 free AI requests per day with no signup and no AI subscription.

What DocPivot AI PDF Editor Does

The editor turns one sentence into an ordered list of editing steps, then runs those steps on your PDF. It supports 21 step types, so a single request can delete pages, fill a form, redact names, and lock the result with a password. You can also skip the AI and build the list by hand from a menu.

• Page edits: delete, keep, rotate, reorder, insert blank pages, and remove blank pages.
• Stamps: text stamps, page numbers, headers and footers, and watermarks.
• Page layout: crop, resize, and N-up, which prints several pages on one sheet.
• Forms: fill text fields, dropdowns, checkboxes, and radio buttons, with optional field locking.
• Privacy and scans: redaction by word or by data type, plus OCR for scanned pages.
• Output: split, compress, and password protect.
• Document AI: translate the document into another language or summarize it.

The main difference from most AI PDF tools is where the work happens. A typical AI assistant uploads your document so a language model can read it. Here, the planner sees only your request and a few facts about the file, while open source libraries (pdf.js and pdf-lib) do the editing on your device. If you would rather click than type, the manual PDF Editor lets you make each change by hand.

How DocPivot AI PDF Editor Works

1. Describe or build. Upload a PDF and type your request in any language, or add steps manually from the menu.
2. Plan. The AI (Google Gemini 2.5 Flash, with automatic fallback models) receives your sentence, the page count, which pages are landscape or scanned, and the names of any form fields. It returns a step list, a short explanation in your language, and a "Not possible here" note for anything it cannot do.
3. Validate. Each suggested step is checked against the real document. A step aimed at a page that does not exist is dropped and reported, form values must match real fields and options, and a password step waits for you to type the password yourself.
4. Review. Change, reorder, or delete any step before anything touches the file.
5. Apply. Steps run one at a time in a fixed order. OCR runs first, then redaction, then page edits, then stamps and form filling. Translation and summaries read the finished text, and compression and password protection run last on the finished file.
6. Download. A result card shows an edited PDF, a ZIP of parts if you split, a translated PDF with a .txt copy, or a summary you can copy or save as .txt. Nothing downloads until you press Download.

If any step fails, the whole run rolls back and your file stays exactly as it was. After a successful run, one Undo restores the original, which makes it safe to try a long request and adjust it.

What You Can Ask DocPivot AI PDF Editor to Do

Requests work best when they name the pages and the action, but the planner also understands page groups such as "the landscape pages," "odd pages," or "the last page." The table below shows typical requests and what happens with each one.

A request can mix any of these, as long as it stays within the combination rules covered further down. When you need only one action, the single-purpose tools open faster. Redact PDF handles a quick redaction on its own, and OCR PDF turns a scan into searchable text without any other edits.

Key Benefits of DocPivot AI PDF Editor

• No AI subscription: The planner is free within the daily allowance of 25 requests, with no account, no trial, and no paid add-on.
• Your file stays out of the AI: Planning works from the page count, page shapes, and form field names, never from the document itself.
• Page numbers mean what you see: "Delete page 2 and rotate page 5" rotates the page shown as 5 in the preview, even though page 2 is already gone by then.
• Bad steps are caught early: Steps aimed at missing pages or form options that do not exist are removed and reported instead of guessed.
• One request replaces several tools: A job that once meant separate trips to Watermark PDF and Add Page Numbers to PDF now finishes in a single pass.
• Works in any language: Requests, explanations, translations, and summaries follow the language you write in.
• Failed runs change nothing: Each run is all or nothing, and Undo restores the original after a successful run.

How DocPivot AI PDF Editor Keeps Your File Private

The planner never receives the PDF or its text. It sees only four things: your sentence, the page count, which pages are landscape or scanned, and the names of form fields. That is enough to build a correct step list, and it means a contract or medical record is never read by a language model during planning.

Most editing happens in your browser. Page edits, stamps, form filling, redaction, and splitting all run on your device. Three steps need the server: OCR (Tesseract), compression (qpdf with image re-encoding), and password protection (qpdf with AES-256). Each server step is labeled on screen, and the file is deleted from the server within 30 minutes.

Your document text goes to the AI only when you add Translate or Summarize, because those steps must read the words. The page says so before you start, so you can leave those two steps out of any run on a confidential file. API keys are stored in the site settings database and never appear in the page code.

When to Use DocPivot AI PDF Editor

Use DocPivot AI PDF Editor when a job needs several actions on one PDF, or when describing the job is faster than hunting for each menu. A single task, such as dragging thumbnails into a new order, can be quicker in Organize PDF.

• Contract preparation: A paralegal deletes a cover sheet, redacts the client name and every phone number, adds a CONFIDENTIAL header, and protects the file.
• Scanned paperwork: An office manager makes a scanned invoice searchable, removes blank pages, and shrinks it for email in one run.
• Repeat form filling: An HR coordinator fills an onboarding form with a name, country, and plan choice, then locks the fields before sending it.
• Multilingual teams: A manager translates a policy into Arabic or Japanese for a regional office and receives a properly set PDF plus a text copy.
• Long reports: A researcher asks for a detailed summary of a 90-page report before deciding whether to read it in full.
• Print handouts: A teacher puts four pages on each sheet with N-up and adds page numbers before printing a class pack.
• Upload size limits: A job applicant crops scanned certificates and compresses the result to meet a portal limit. If size is the only problem, Compress PDF does that alone.

DocPivot AI PDF Editor Compared With Adobe Acrobat AI Assistant and PDFKro

Adobe Acrobat AI Assistant and PDFKro both edit PDFs from typed commands, but they differ in cost, scope, and where your file goes. The details below reflect each product's public pages as of October 2026.

Each tool wins in a different place. Adobe is the stronger choice when you need to rewrite text inside the PDF or want a translation that keeps the original layout. PDFKro handles merging several files in the same command. DocPivot fits when you want redaction, OCR, form filling, and translation in one free run, with page edits handled in your browser and no file sent to the AI for planning.

DocPivot AI PDF Editor Limits and Known Gaps

The editor is built for one document and a defined set of actions, so some jobs belong in other tools.

• One PDF per run: The editor accepts a single PDF. Combine files first with Merge PDF, then edit the result.
• No text rewriting: The AI can stamp, redact, and fill fields, but it does not change existing sentences. For that, use Edit PDF Text.
• No signatures: Signing is not one of the 21 step types. Sign PDF handles that job.
• Split combinations: Split cannot share a run with compress, protect, or translate, and the step list explains why. Compress first, then run Split PDF on the smaller file.
• Redaction side effects: Redaction turns affected pages into images, so those pages cannot be summarized or translated later in the same run. Rotated pages are skipped, and the result tells you how many.
• Translation layout: A translation is a clean, reflowed document with headings and paragraphs, not a copy of the original layout. Arabic, Hebrew, Persian, and Urdu are set right to left, and Chinese and Japanese wrap by character.
• Summary length: Summaries read up to about 120,000 characters, and the result card says so when a document is longer.
• Daily allowance: Each visitor gets 25 AI requests per day, and a translation counts as one request for roughly every 40 pages.

Tips for Writing Better Requests in DocPivot AI PDF Editor

Clear requests produce clean step lists on the first try, which saves both time and daily requests.

• Use preview page numbers: Refer to pages as they appear in the preview, and the tool keeps them pinned even after deletions.
• Do not worry about order: Write steps in any order, because OCR, redaction, page edits, stamps, and output steps always run in their safe sequence.
• Match form wording: Use the exact option text from a dropdown, such as "Pro plan," so validation accepts the value.
• Combine names and data types: A request like "redact Maria Lopez, all phone numbers, and all IBANs" covers both specific words and patterns in one step.
• Read the "Not possible here" note: It lists anything the planner skipped, so you know what to handle elsewhere.
• Keep sensitive text local: Leave Translate and Summarize out of runs on confidential files, since those are the only steps that send text to the AI.

DocPivot AI PDF Editor FAQ