Guides

Complete workflows, start to finish.

8guides that cover nearly every document you'll meet. Each one explains the situation it is for, the steps in order, and the mistake people actually make. For quick answers instead, see the mini tutorials or the FAQ.

Last reviewed: August 8, 2026

Use the PDF power tools

Transform pages or document security without sending the file to a server.

The editor is built for filling and marking up a single document. When the job is structural — combining several files, pulling out a page range, changing page order, turning a scan upright, numbering a finished packet, or adding and removing encryption — a focused tool is faster and safer, because each one does exactly one thing and shows you the result before anything downloads.

  1. 1

    Open Tools and choose the focused workflow: Merge, Split, Organize, Rotate, Page Numbers, Protect, or Unlock.

  2. 2

    Choose PDF files and review page counts, thumbnails, encryption status, and operation settings.

  3. 3

    For protected inputs, enter an authorized password in the password field shown for that file.

  4. 4

    Set file order, page ranges, page selections, numbering, or permission options as required by the tool.

  5. 5

    Generate the result. Processing runs in a browser worker and the output downloads only after validation succeeds.

  6. 6

    Reopen important output to confirm page order and security before sharing or submitting it.

Watch out: Order of operations matters more than most people expect. Unlock before you edit, edit before you number, and protect last — otherwise numbering lands on pages you later reorder, or encryption blocks the next tool in the chain.

Fill a government form

The bread-and-butter workflow: a dense official form with small boxes and strict layout.

Official forms are the hardest documents to fill by hand and the ones where a mistake costs the most time. They pack many small boxes into a fixed layout, they often arrive already part-completed by an office, and they are frequently the exact documents you would rather not upload to a stranger's server. Running the whole job in the tab solves the last problem outright and the first one with field detection.

  1. 1

    Upload the PDF from the editor's start screen — the AI scan runs automatically when the document opens.

  2. 2

    Review the detected fields. Fields that came with the form (even locked, completed ones) arrive pre-filled with their existing values.

  3. 3

    Click any field to fill it. The fill bar keeps focus so you can Tab through the document top to bottom.

  4. 4

    Toggle checkboxes with a single click; use the cross tool for boxes that expect an X instead of a checkmark.

  5. 5

    Zoom in for tight rows and use the numeric position controls to nudge a field into exact alignment.

  6. 6

    Export when done, then open the file once in a PDF viewer to confirm every value landed where it should.

Watch out: Check the boxes the scan filled in for you before you check the ones it left empty. A field that arrived with an existing value is showing you what the form already says, not what you typed — reading past those is the most common source of a rejected submission.

Sign and export cleanly

Add a real drawn signature and produce a submission-ready file.

A signature drawn now is represented as normalized vector strokes, so it has no background rectangle and stays sharp when resized. Legacy transparent bitmap signatures remain supported for compatibility. Typed text goes out in one of the fonts every PDF reader is required to support, so the file renders the same for whoever opens it.

  1. 1

    Select the signature tool and click where the signature belongs — a drawing pad opens.

  2. 2

    Draw with a mouse, finger, or stylus. Redraw as many times as you like before accepting.

  3. 3

    Resize the placed signature to fit the line; the stroke stays sharp at any size.

  4. 4

    Add the date beside it with a text or number field using an export-safe font.

  5. 5

    Export to download, or print directly, then open the result once to confirm it looks the way you expect.

Watch out: The export is vector wherever it can be. If the source document cannot be decrypted or its internal structure is damaged, the export falls back to rebuilding pages as images so the result is still correct — accurate on screen and on paper, but no longer selectable text. Open the finished file once and try to select a word if searchable text matters to you.

Erase and highlight cleanly

Both tools are text-aware: they follow the words underneath instead of guessing.

A freehand rectangle over text always looks freehand. Because both tools read the text layer beneath the cursor, a highlight snaps to the lines it actually covers and an erase box hugs the words you dragged over, so a marked-up page reads as deliberate rather than patched. That also means they behave differently on a photograph of paper, where there is no text layer to follow.

  1. 1

    Pick the highlight tool and drag across the passage — one marker stroke snaps to each covered line, blending like a real highlighter.

  2. 2

    Click a single line with the highlight tool to mark the whole visible line in one go.

  3. 3

    Pick the erase tool and drag over unwanted text — the cover box hugs exactly what you covered.

  4. 4

    Match the erase color to the page background with the swatches if the paper isn't pure white.

  5. 5

    Adjust any placed highlight or erase box afterwards: move it, resize it, or delete it like any other element.

  6. 6

    Choose Cover for correction-fluid cleanup. Choose Remove only when you accept a rasterized page and its lost text, links, fields, annotations, and tags; then inspect the export before sharing it.

Watch out: Cover mode paints over the words while leaving the underlying PDF content recoverable. Remove mode rasterizes each affected page so the covered content is absent from the export, but it also removes selectable text, links, form fields, annotations, and tags on that page. Neither mode is a certified redaction workflow, so read the redaction guide and verify the finished file before sharing it.

When the AI scan misses fields

No detector is perfect. Here's the fast path from a noisy scan to a tidy form.

The scan is a local detector, not a cloud model reading your page. It works from what the document actually contains — real form fields where they exist, drawn boxes and rules, and text labels — so its accuracy tracks how much structure the file carries. A digital form built for filling gives it plenty; a photograph of a printed page gives it almost nothing.

  1. 1

    Run Rescan first — it merges newly found fields without duplicating the ones you already have.

  2. 2

    Missing a field? Add it manually with the text, number, check, or signature tools; manual fields behave exactly like scanned ones.

  3. 3

    Got a duplicate or misplaced box? Select it and delete or drag it — AI fields are never locked.

  4. 4

    If the scan is mostly noise on an unusual document, use “Clear AI fields” and place fields manually — your own work is kept.

  5. 5

    Scanned-image PDFs (photos of paper) have no machine-readable text; manual tools are the right approach there.

Watch out: One bad box is faster to delete than to rescan away. Rescanning is for documents whose structure changed; it is not an undo for a single misplaced field, and running it repeatedly on the same page will not produce a different answer.

Install it and work offline

Set the app up once while you are connected so a lost signal stops nothing.

Every document operation already runs on your device, so the network is only ever delivering the application itself. Installing Novus registers the site with your operating system as a standalone app and lets the browser keep the app's own files locally — which is why a merge, a fill-and-sign, or an export still finishes in airplane mode. It is also the cheapest privacy check available: work that completes with the network off was never being sent anywhere.

  1. 1

    On Chrome, Edge, or Android, accept the install prompt or use the browser menu's Install entry.

  2. 2

    On iPhone or iPad, open the site in Safari and choose Share, then Add to Home Screen.

  3. 3

    Open the editor once with a real PDF while you are still connected so its rendering engine is on the device.

  4. 4

    Visit the tool pages you expect to need — a single visit is enough to keep each one available.

  5. 5

    Copy the source documents onto the device rather than leaving them in cloud storage you will not be able to reach.

  6. 6

    Rehearse: switch on airplane mode and run one small export before you actually need it to work.

Watch out: The first visit has to be online, and a page you have never opened may not be available offline. Advertising and analytics requests are deliberately never cached, so those areas stay blank without a connection — that is expected, not a broken page.

Create a PDF from a blank page

The one workflow here that starts with no file at all: choose the paper, stack the blocks, download a real PDF.

Every other surface on this site opens a document you already have. The creation zone opens nothing. A document here is an ordered stack of blocks — headings, paragraphs, images, tables, shapes, rules, spacing and page breaks — flowed down the page until the room runs out, at which point the next page begins. Knowing that before you start typing explains both what is easy (rewriting, restyling or reordering one block without disturbing anything below it) and what is not (dropping something at an exact spot on the paper, which is a job for the editor once the file exists — the shape block is the one exception, and it takes coordinates rather than alignment).

  1. 1

    Open Create and choose one of the eight starting points. Picking one replaces the title, file name, page setup and every block below it, so choose before you write rather than after.

  2. 2

    Set the paper up: A4, Letter, Legal, A3, A5, Tabloid, or a custom size typed in points, portrait or landscape. Margins come as narrow, normal and wide presets or as four numbers you type yourself. 72 points is one inch.

  3. 3

    Build the document from the block buttons — heading, paragraph, list, image, table, shape, divider, spacing and page break. Each card carries arrows to move or delete it, content changes have undo and redo, and everything below reflows.

  4. 4

    Reach for the table block whenever something has to line up in columns, and the shape block when it does not. Rectangles, ellipses, lines and arrows use labelled x, y, width and height controls, so positioning and resizing never require dragging. A negative vertical offset pulls a shape back over earlier blocks, while graphics remain behind the text.

  5. 5

    Style each text block on its own: Helvetica, Times or Courier, a size in points, a colour, left, centre or right alignment, bold and italic. Headings come in three levels at 24, 18 and 15 points, bold until you say otherwise.

  6. 6

    Read the preview as you work. It is not a second renderer — it is the exporter's own layout drawn to the screen with the same embedded fonts, and the dashed rectangle on each page shows exactly where your margins fall.

  7. 7

    Fill in the title, author and file name. The first two are written into the PDF's metadata; the third names the download, with “.pdf” added for you.

  8. 8

    Download the PDF and open it once. The text is real text, so a search inside your PDF viewer should find a word you typed.

Watch out: Nothing is saved. There is no account, no server and no autosaved draft — the document lives in this browser tab and closing it loses everything you typed. Download the PDF as soon as it is worth keeping, and treat the exported file as the copy of record.

Make a form people can fill in

Write the form in the creation zone, export it, then let the editor's field scan find the lines you ruled.

A form is the one document where creating and filling meet. The creation zone rules the answer lines; the editor's scan reads them back. That scan is rule-based rather than a model call — it merges the form fields already embedded in a PDF, the boxes and rules drawn into the page, and the labels sitting in the text — and a long run of underscores is one of the shapes its text pass looks for. That is why the intake-form template rules its answers with underscores inside a Courier block instead of with decorative lines: the rule is text, and text is what comes back as a field.

  1. 1

    Start from the intake-form template, or add your own paragraph, set it to Courier, and rule each answer line with underscores.

  2. 2

    Do not try to line the rules up by padding with spaces. A paragraph collapses every run of whitespace to a single space before it is measured, so the padding is gone by the time the font matters and a monospaced face does not rescue it. The run of underscores survives — it is one unbroken word — so the scan still finds each rule; the rules simply start wherever their label ends.

  3. 3

    If you want the rules in a straight column, put the form in a table block instead: labels in the first column, a run of underscores in the second, borders set to none. Every cell in a column starts at the same x, and the underscores still come back from the scan as fields.

  4. 4

    Rule the signature line the same way and label it — the scan reads “Signature” beside a rule and places a signature field rather than a plain text box.

  5. 5

    Export the PDF, then open that file in the editor. The scan runs automatically as the document opens.

  6. 6

    Review what it placed: move, resize or delete anything that landed oddly, and add what it missed with the text, number, check or signature tools.

  7. 7

    Fill it in and sign it, then export again. What comes out is an ordinary PDF, so it can still be combined, numbered or password-protected afterwards.

Watch out: A scan is a starting point, not a promise. Check the placed fields before anyone types into them — every field, scanned or added by hand, can be moved, resized or deleted, and a blank the scan did not find is faster to place yourself than to argue with.