Document Upload
Document upload lets you add files directly to KnowStack for Knowledge Base generation. This is useful for internal documents, reports, manuals, and other content that is not available via web crawling or email.
Supported File Types
| Format | Extension | Max Size | Notes |
|---|---|---|---|
| 25 MB | The PDF's own text is read first. Scanned PDFs without a text layer are read with OCR. | ||
| Word | .docx | 25 MB | Microsoft Word documents. Formatting is stripped; text is preserved. |
| Excel | .xlsx | 25 MB | Each sheet becomes a table, up to 1,000 rows per sheet. The older .xls format is not accepted: save it as .xlsx first. |
| Images | .png, .jpg, .jpeg, .webp | 25 MB | Read by an AI vision model, for example photos of printed pages or screenshots. |
| Plain Text | .txt | 25 MB | Raw text files. |
| CSV | .csv | 25 MB | Comma-separated values, useful for structured data. |
How to Upload Documents
- In the sidebar, open 'Sources', then 'Documents'.
- Drag and drop files onto the upload area, or click 'Drop files here or click to browse' to open the file picker.
- You can select multiple files at once. If a folder is open, the files are added to it.
- The upload starts immediately.
- Once uploaded, text is automatically extracted from each file in the background.
- Uploaded documents appear in the list below.
Viewing and Managing Documents
- Preview -- Click on any uploaded document to preview its extracted text in a modal
- Delete -- Remove individual documents you no longer need
- Delete All -- Remove all uploaded documents at once
- Extraction status -- Each document shows 'Pending', 'Processing', 'Ready', 'No text' or 'Failed'
Text PDFs give the most accurate result. A scanned PDF that has no text layer at all is read with OCR automatically. A PDF that has text on some pages and scanned images on others is read from its text layer only, so the scanned pages can be missed. Extracting text from PDFs and images counts toward your AI usage.