Seventy-plus tools across seven categories. Here's what each one actually does, when you'd use it, and which ones run in your browser vs. on our servers.
Arawa PDF ships with seventy-plus tools at launch, grouped into seven categories. Some run entirely in your browser. Some need server-side engines (Ghostscript, LibreOffice headless, Tesseract OCR, Chromium for HTML, AI models for summarisation). This post walks through every one of them.
Tools for rearranging a PDF without touching its content.
• Merge PDF — combine multiple PDFs into one, in any order. Server-side.
• Split PDF — break a PDF into smaller files by page range or every-N-pages. Server-side.
• Organise PDF — reorder, duplicate, and delete pages with drag-and-drop. Client-side for the layout, server-side to write the final file.
• Remove pages — drop selected pages and re-output. Server-side.
• Extract pages — pull a subset out into a new file. Server-side.
• Rotate PDF — rotate selected pages by 90, 180, or 270 degrees. Server-side, fast.
• Crop PDF — trim margins or focus on a region. Server-side.
Tools for making a PDF smaller, repaired, or archival-grade.
• Compress PDF — three quality levels. Targets common sources of bloat (uncompressed images, embedded fonts). Server-side via Ghostscript.
• Repair PDF — fix structurally broken files that other readers refuse to open. Server-side via pikepdf.
• OCR PDF — make scanned (image-only) PDFs searchable and copyable. Server-side via Tesseract via ocrmypdf. Supports the languages listed in our language picker.
• PDF to PDF/A — convert to the long-term-archival PDF/A standard. Server-side via Ghostscript.
Tools for getting other formats into PDF.
• Word to PDF — .doc and .docx via LibreOffice headless.
• Excel to PDF — .xls and .xlsx via LibreOffice headless.
• PowerPoint to PDF — .ppt and .pptx via LibreOffice headless.
• JPG to PDF — wrap one or more images into a single PDF. Server-side.
• HTML to PDF — supply a URL or paste HTML; we render via Chromium. Useful for receipts, invoices, and saved web pages.
Tools for extracting structured content out of a PDF.
• PDF to Word — best-effort text and layout extraction. Server-side.
• PDF to Excel — table detection and extraction into spreadsheets. Server-side.
• PDF to PowerPoint — one slide per page. Server-side.
• PDF to JPG — rasterise pages to images at chosen DPI. Server-side via Poppler.
Tools for changing what's on the page.
• Edit PDF — add text, shapes, and images to existing pages. Runs in your browser; we only write the final file server-side.
• Add watermark — text or image, configurable position and opacity. Server-side.
• Add page numbers — multiple position and format options. Server-side.
• Compare PDFs — diff two PDFs side-by-side. Fully client-side; the files never leave your browser.
• Redact PDF — black out text or regions. Client-side for the selection, server-side to permanently flatten.
• Fill PDF forms — type into AcroForm fields and save the result. Server-side via pikepdf.
Tools for locking and unlocking PDFs.
• Protect PDF — add an open-password and permission flags (printing, copying, editing). Server-side via QPDF.
• Unlock PDF — remove a password you already know. We do not crack passwords; you must supply the correct one.
• Sign PDF — place a visible electronic signature with an audit trail. Combines client-side placement with server-side flattening and audit-log recording.
Tools powered by large language models, available on Premium.
• Summarise PDF — short, medium, or long summaries of the document text. Sends extracted text (not the file itself, not images) to Anthropic's Claude API. Output stays in your account session and is deleted on the same one-hour schedule as the source file.
• Translate PDF — translate the document into any of the supported languages, preserving rough structure. Same processing model as Summarise.
More on how these work in a separate post.
A few tools you might expect that we have decided against, on purpose:
• We don't "unlock" passwords we don't have. If you've forgotten the password, no honest service can recover it.
• We don't run a cloud-import flow yet (Google Drive, Dropbox). It would require holding OAuth tokens for those providers; we'd rather not until we have a clean design for the security review.
• We don't keep version history. Once a file is processed, the previous version is gone with the next sweep.
You can take this list as a roadmap. Things that fit our design constraints will eventually arrive.