Skip to main content

File Uploads

SVAHNAR provides a streamlined pipeline to ingest PDF documents into your Knowledge Repository so your agent can read, analyze and answer queries using the document contents (RAG-style).

Key capabilities​

  • Single or bulk PDF upload — Upload one or many PDF files in a single operation. The system queues files, extracts text, and indexes content into the repo.
  • Built‑in OCR — If a PDF contains scanned images (pictures of text), SVAHNAR runs OCR automatically and extracts readable text so nothing is left behind.
  • RAG-ready knowledge store — Uploaded documents are parsed, chunked, and embedded so that retrieval-augmented generation (RAG) can surface exact, context-aware snippets to the agent.
  • Agent access & analysis — Your agent can access the indexed documents for deeper analysis, summarization, question-answering and contextual reasoning across all uploaded files.
  • Searchable knowledge — Once ingested, documents become queryable: ask natural-language questions and receive answers that reference the original document passages.

Typical workflow​

  1. User selects PDFs (single or multiple) via the Upload UI or API endpoint.
  2. Server receives files and performs text extraction. If pages are images, the OCR sub‑pipeline runs automatically.
  3. Extracted text is normalized and split into chunks for indexing.
  4. Chunks are embedded and stored in the Knowledge Repository.
  5. Agent queries the repository; retrieved context is used to generate accurate, source‑backed responses.