Document Upload

If you already have documents, you can upload them directly instead of rewriting them one by one. HaveAI automatically processes the file, extracts the text, splits it into chunks and adds it to your assistant's knowledge base.

Supported Formats

HaveAI can process a wide range of document and text formats:

  • Documents: PDF, DOCX, DOC
  • Text: TXT, MD (Markdown)
  • Size limit: up to 25Β MB per file

Text extraction is essential

HaveAI extracts text from documents. So scanned (image-based) PDFs or image-only documents may not give good results. Text-based documents with selectable content yield the best outcome.

Uploading a Document

1

Go to the document section

Open AI Training β†’ Document Upload in the panel.
2

Select or drag a file

Choose the document to upload. Processing starts automatically once the upload completes.
3

Wait for processing

The document goes through text extraction and then the embedding (vectorization) stage. The status updates from processing β†’ completed. Once completed, the assistant can use the document's content.

Processing Stages

Every uploaded document goes through two stages:

  • Text extraction: Text is extracted based on the document format.
  • Embedding: Text is split into chunks and vectorized for semantic search.

You can see how many chunks each document was split into and its training status in the document list.

Well-structured documents give better results

Documents with headings, lists and clear paragraphs produce more accurate answers than messy or single-block documents.

Secret keys are automatically redacted

If your uploaded documents contain sensitive data such as API keys, passwords, or private keys, HaveAI automatically detects and masks them before the content is processed. Secrets are never stored in the database or exposed in assistant replies β€” you are safe even if they are uploaded by mistake.

Related Topics

Was this page helpful?