Documents are the knowledge base for your agents. Upload files to a datastore and Orka handles chunking, embedding, and indexing.
| Format | Extension |
|---|
| PDF | .pdf |
| Word | .docx |
| Plain Text | .txt |
| Markdown | .md |
Upload Documents
- Open a datastore from the Datastores page
- Click Upload or drag and drop files
- Documents are processed automatically
You can upload multiple files at once. Each file becomes a separate document in the datastore.
Processing Pipeline
When you upload a document, Orka:
- Extracts text content from the file
- Chunks the text into semantic segments
- Embeds each chunk using a vector embedding model
- Indexes the embeddings for fast retrieval
Document Status
Each document shows a processing status:
- Processing — Currently being chunked and indexed
- Ready — Available for agent queries
- Failed — An error occurred during processing
Manage Documents
From the datastore detail page, you can:
- View document metadata and chunk counts
- Delete individual documents
- Re-upload updated versions
