Compare the text with its source
Read the original beside the extracted text. Correct a section before it becomes part of your knowledge base.
Open source comparisonLoading Precidoc…
DOCUMENT PREPARATION FOR AI · EARLY ACCESS
Better documents.
Better AI outcomes.
Messy extraction and missing context make documents harder to use. Extract text from PDF, DOCX, or Markdown, fix what needs attention, and export reviewed chunks with their sources.
Processed in your browserNo API key required
Real browser parsing and rule-based checks. No signup for the sample. LLM assistance is planned; it is not used today.
Check the text before you use it.Compare sections, edit the output, then export.
STRUCTURED CONTENT. CONNECTED SOURCES.Explore each step
Readiness from this sample78/100Review required
Original text
section_03## Recovery procedure 1. Check replication status before recovery. 2. Verify replication lag and available disk space. 3. Restore using the approved playbook. | Parameter | Limit | | --- | --- | | Replication lag | 30 seconds | | Free disk | 20% | Record verification results and execution time in the incident record.
3 document sections
Extracted text, organised by section.
section_03SourceExtract
Select a section to see its source text.
Paused
INPUT FORMATS
Digital PDFDOCXTXTMarkdown
Export as JSON or MarkdownDOCUMENT REVIEW
Check the text against the source.
This example contains 3 sections and 1 table. Open it in the sample workspace to compare the text with its source.
section_03Extracted text
Recovery procedure
- Check replication status before recovery.
- Verify replication lag and available disk space.
- Restore using the approved playbook.
| Parameter | Limit |
|---|---|
| Replication lag | 30 seconds |
| Free disk | 20% |
- Version
- Not set
- Tables to review
- 1
- Source sections
- 3
From the built-in sample runbookOpen sample
See how the text is split
Inspect each chunk and its page or section reference. Adjust the size and merge repeated sections in your workspace.
Inspect the chunksDownload the reviewed document
Confirm your review, then export JSON or Markdown with the document metadata and source references.
See the exportCommon questions
What works today, and what needs your review.
Which documents are supported?
Digital PDFs, DOCX, TXT, and Markdown. The trial allows one document up to 5 MB and 20 PDF pages. Registered workspaces support 10 documents per tab, 15 MB per file, and 200 PDF pages. Scanned PDFs still require OCR, which is not available.
Are my files uploaded to a server?
No. Files are processed in your browser and kept temporarily in the tab’s memory. Reloading or closing the tab clears the session, so export your results first.
Does Precidoc use AI?
The current workflow uses extraction and rule-based checks. LLM processing, OCR, embeddings, and direct knowledge base connections are not available yet.
Can I use the output right away?
The output is a preparation package. Verify tables, numbers, and PDF reading order, then adapt the package to the format your system requires.
Try your own document.
Open the workspace, add a file, and inspect the extracted text.