Frequently Asked Questions

Core ExactPDF browser tools are free. Max is our paid document-intelligence service, with web subscriptions and API/MCP credit options. Prices, included work and current workflow limits are shown before purchase.

Browser-local tools process files on your device. Hosted workflows, including Max and Office conversion, upload the selected file for server processing. Review the workflow's data handling and retention before submitting sensitive documents; local-processing claims do not apply to hosted jobs.

We use browser security headers and validation and rate limits on selected routes. Hosted Max jobs require authentication, and job status and downloads check the owning account. Signed download links expire, but anyone holding a valid link may access the file. These controls reduce specific risks; they are not a security or compliance certification and do not make every document safe to upload.

The tools directory lists browser workflows for compression, merging, splitting, viewing, editing, OCR, redaction, conversion, Markdown and more. Max adds hosted compression, folder processing, narration, source packs and reviewed redaction through Web, MCP or API. Availability, input limits and output quality depend on the specific workflow; a tool name is not a guarantee that every PDF or language is supported.

Client-side processing means the heavy lifting is done by your own computer's processor (CPU) rather than a remote cloud server. This ensures your data stays on your machine and often results in faster processing for small to medium-sized files.

Many browser-local tools can continue after their required engine resources have been prepared while connected. This is not a cold-start guarantee: offline availability depends on the tool and on your browser retaining its downloaded resources.

Browser-local tools do not send your document bytes to a processing server. Downloaded outputs remain on your device under your control. Office/Word to PDF is a clearly marked hosted workflow: it sends the selected document to an isolated LibreOffice job and attempts to remove its temporary input and output after the final conversion request. Interrupted uploads are eligible for stale-job cleanup after 30 minutes, triggered by a later request; this is not a guaranteed deletion deadline. Max and API are explicit hosted services with separate storage and retention controls.

Background-job input, intermediate and output files in Max hot storage are configured for deletion after one day. Lifecycle deletion runs asynchronously, and deleted objects remain recoverable through seven-day soft delete. This is not guaranteed permanent erasure at 24 hours. Signed download links expire separately. Account, billing and job metadata have separate retention and are not deleted by the file lifecycle. Self-serve deletion, retention choices and deletion receipts are not currently available; do not upload documents that require those controls.

Max is one document-intelligence product with web and automation access. Currently, web subscriptions and API/MCP credit balances are billed separately and are not interchangeable. Check the workflow and allowance included in your selected option before purchasing.

PDF compression, Folder Studio, MP3/M4B/MP4 narration, verified source packs and Smart Redaction are available through the web workspace and paid MCP/API. Compression accepts up to 1 GiB per PDF; Folder Studio accepts a PDF-only ZIP up to 256 MiB; asynchronous source packs accept text PDFs up to 256 MiB and 500 pages; Smart Redaction accepts up to 256 MiB and 100 pages. Its inspect and reviewed apply steps are separate jobs: API/MCP charges one credit for each verified output, while web uses one workflow allowance for each. Workflow-specific page, output and monthly allowances are shown before starting. Narration requires English embedded text and accepts PDFs up to 10 MiB, with up to 20 selected pages, 18,000 characters, 20 estimated minutes and one active job. Translation is not available. The synchronous source-pack endpoint has separate, smaller limits; check the API reference for the endpoint you use.

The shared document-workflow API reserves your approved credit budget before processing. A completed batch charges only for successfully processed PDFs and returns unused reserved credits. Failed jobs return their reserved credits automatically; Web jobs return their reserved workflow allowance. Narration keeps its separate per-minute pricing and failure-refund policy. Shared document-job creation uses an Idempotency-Key: reuse the same key and request when recovering that submission, and check its job status before creating another. A new submission or a different endpoint is not universally duplicate-safe; do not automatically retry ambiguous paid requests.

Automatic mode uses editable text for text-only PDFs and page images for illustrated documents. Preserve appearance keeps the visible page layout, artwork and tables as images in a Word document, up to 50 pages; the text and table cells in those images are not editable. Editable text extracts words but does not reconstruct artwork, tables or page layout. English OCR is a separate editable-text option for scans, limited to 15 pages. Review the downloaded Word document before relying on it.

No. Find exact text searches for one actual value or phrase, such as Ada Example or TEST-4821; it does not interpret Name, ID as field categories, separate search terms or instructions to redact their values. Search each actual value separately or draw manual redaction boxes, then review every page. Find searches selectable text first and uses English OCR when no selectable match is found on a page. It covers matching text runs, which can also hide nearby words. Automatic local detection suggests emails, US-style phone numbers, SSNs and card-like numbers; it does not reliably identify arbitrary names or ID types. OCR may miss text, particularly in scans or unusual layouts. QR codes, barcodes and photos need manual review and redaction; their encoded contents are not automatically removed. Max also requires you to review detected regions before applying redaction.

Free Smart Redaction rebuilds the reviewed page images into an image-only PDF, then checks the exported artifact. The original text layer, attachments, scripts, forms, comments and hidden layers are not copied. The local workflow reviews up to 10 pages; if a file exceeds 10 pages or 5 MiB, it warns that only the reviewed first pages will be exported. Unreviewed pages are excluded, not certified as redacted. Approved search values are rescanned across every exported page; unavailable extraction or OCR blocks verification. Verification checks the marked regions and available approved-term evidence, not whether every sensitive item was detected. OCR can miss visible or encoded information. Image-only output loses searchable text and interactive features. Review the exact downloaded PDF; the report is not a legal certificate.

No. Size targets are best-effort. Results depend on the source PDF and the quality trade-offs it permits. Check the measured output size, readability, forms and signatures before submitting it. If a target cannot be reached acceptably, use a different target or split the document rather than assuming that a successful download meets the limit.

We use Google Analytics 4 and Ahrefs Web Analytics to understand aggregate page and tool usage. GA4 is configured without Google Signals or ad-personalization signals. We do NOT use Facebook Pixel, remarketing pixels, or affiliate tags. File names, PDF contents, extracted text, and arbitrary query parameters are never part of analytics telemetry. You can inspect these requests in Developer Tools or block analytics without breaking the tools.

Open Developer Tools (F12) → Network and process a PDF in a browser-local tool. You will see page assets and analytics requests, but no request containing your PDF bytes. Some tools can continue after their required resources are prepared; server-backed tools, including Office to PDF and paid Max/API workflows, display their processing model separately.

ExactPDF is part of Aurabridge Systems. Aurabridge Systems owns the ExactPDF brand and its associated intellectual-property rights. Third-party trademarks and open-source components remain subject to their respective owners' rights and licences.

If you find our tools useful, you can make a voluntary contribution via 'Buy me a coffee' or simply share the tool with your friends and colleagues!