From your document to a usable answer
- 1
Check for selectable text
If your PDF is a scan or photograph, Adaped cannot read it. Start from the original Word or text document if you have it. Do not assume a page image will be anonymised.
- 2
Read the extracted text
Add the PDF, then check the reading order. Columns, page breaks and tables can change how the text appears. Confirm that the content still supports the question you want to ask.
- 3
Replace selected details
Highlight the original name or phrase. Check for split words, initials and alternate spellings separately. Download a fresh Word or text copy, never the original PDF, for your AI.
- 4
Return the answer
Ask the AI to keep complete REF_ labels. Add its answer in Restore answer at the same address and browser profile, or paste the text, then download the restored result.
A separate text copy, not marks over a PDF
Adaped rebuilds the output from the text you review. It does not cover text with coloured rectangles or preserve the original PDF binary. Images, annotations, attachments and original metadata are not copied to the prepared file.
That makes the output suitable for a text-based AI task when the extraction is complete enough for your purpose. It does not produce a court-ready, publishable or layout-preserving redacted PDF. Use a specialist PDF redaction process if that is what you need.
Check what extraction leaves out
PDF form fields are not imported. Multi-column reading order, unusual fonts and text broken across lines can affect the extracted wording. A phrase that is split by a line break may need a separate selection.
Import stops if a page has no extractable text, rather than silently skipping an image-only page. The current limits are 8 MB, 100 pages and 250,000 extracted characters. There is no OCR or password-protected PDF support.
Keep useful context without unnecessary detail
Select names, organisations, email addresses and private phrases you do not need to share. Read the remaining text too: an unusual role, exact date or event can identify someone even when their name has gone.
Only the prepared copy belongs in your AI account. The original PDF, any optional restore file and the restored answer can contain the real details. Follow your organisation's sharing rules; selected replacement is not a guarantee of anonymity.
Before you start
Can I download an anonymised PDF?
No. PDF input becomes a new text-based .docx or .txt file. The supported job is preparing the extracted wording for AI, not preserving or editing the original PDF pages.
Can I restore a PDF answer?
You can add a returned PDF if its text can be extracted and the complete labels remain intact. Its restored output is Word or text, not PDF.
Updated . Describes the current app; use fictional data first.