Control panel → Import — upload documents for indexing in the knowledge base. Support for many formats and batch upload.
#Supported formats
| Format | Extension | Description |
|---|---|---|
| Documents with a text layer | ||
| Word | .docx | Microsoft Word |
| Text | .txt | Plain text |
| Markdown | .md | Markdown files |
| RTF | .rtf | Rich Text Format |
| ZIP | .zip | An archive of the formats above; images inside it are attached to the articles |
#How to import
- Go to Articles → click Import (or Integrations → the "File import" card)
- Under "Where to import", pick a collection in the current space
- Drag & drop your files or click to select them
- The upload starts right away — watch it in the "Current session" panel
Instead of uploading a file you can paste a link under "Import from URL": Google Docs, Dropbox and direct links to PDF, DOCX, TXT and MD files are supported.
LiKE automatically:
- Extracts text from the document
- Creates an article with a title taken from the file name
- Splits it into chunks for semantic search
- Generates embeddings for AI search
#Status after upload
Imported articles are created as drafts:
- Review the content in the editor
- Edit if needed (title, structure)
- Click "Publish" to index it in search
#Limitations
| Parameter | Value |
|---|---|
| Maximum file size | 20 MB (ZIP archive — 24.8 MB, the most a single upload can carry) |
| Files at a time | Up to 50; a large selection is sent in parts automatically |
| Inside a ZIP archive | Up to 5000 files and up to 500 MB once unpacked |
| PDF without a text layer | Not supported (OCR-processed only) |
| Recommendation for >100 pages | Split into parts |
If a file is not accepted. Sizes are checked before sending: anything over the limit is not uploaded, the rest goes through. For every refusal LiKE says what is wrong and what to do: for a file that is too large — its size and the limit, with advice to split the file or archive into parts; for a damaged archive — recreate it; for an unsupported format — save the file in a supported one. A file that could not be processed stays in the Current session panel with the reason and a Retry button.
#Language
The language of imported documents is detected automatically from their content — you don't need to set it during upload. 25+ languages are supported. The detected language is shown in the article's properties.