How to Rename Scanned PDFs with AI

A normal text PDF contains extractable text. A scanned PDF is often just a page image wrapped inside a PDF container. That distinction matters because the extraction path is different.

• Text PDF: RenameClick can extract text directly.
• Scanned PDF: RenameClick falls back to image-style analysis of the page.

The full behavior is described in Supported File Types. In practice, the scan path is more sensitive to shadows, blur, low contrast, and bad cropping.

1. Start in the Rename workspace with a small batch of scans.
2. Let RenameClick analyze the files and suggest descriptive names.
3. Review outliers before applying anything on disk.
4. Apply approved names, then expand to the next batch.

This is safer than trying to rename 500 scans at once. OCR-heavy workflows always have edge cases, so the goal is fast review, not blind automation.

For the workspace flow itself, see the Rename Workspace documentation.

Scanned PDFs become much more reliable when you stop asking for a generic filename and instead ask for a strict document pattern: document type, date, vendor or patient, and one or two key identifiers.

The Custom Instructions docs already include strong examples for bills and medical documents. The general rule is to make the prompt strict: exact fields, exact order, and explicit fallback text such as UNKNOWN.

When a custom prompt is active, it defines the output format directly. That means the standard format selector and output-language selector stop being the main source of truth for the filename.

After extraction is stable, use Format Patterns to keep names consistent across the whole archive.

For scanned PDFs, file dates are usually more useful than EXIF, because scans typically do not carry camera metadata the way phone photos do. A simple pattern like $date{YYYY-MM-DD}_$1 is often enough to make a scan archive sortable.

If you need invoice-style names, combine strict prompt output with a lightweight pattern instead of trying to cram every field into the prompt and the pattern at the same time.

Most scanned-PDF failures come from one of three things:

• Low-quality scans: blur, skew, dark shadows, poor contrast.
• Ambiguous prompts that do not define the output clearly enough.
• Long documents where only part of the extracted content matters for naming.

RenameClick trims extracted text for long documents, so use prompts that focus on the fields you actually need instead of asking for a full summary. For machine-specific issues or slower local inference, use the troubleshooting guide.