Upload Documents to Knowledge Base

Import PDF, DOCX, TXT, and other files to your AI voice agent's knowledge base for RAG-powered conversations

Upload Documents to Knowledge Base

Import documents from your computer to build your knowledge base. Supported formats include PDF, Word documents, text files, and more.

Upload now: Open your Knowledge Base and click on a knowledge base to upload documents.

Looking to import web content instead? See Web Crawling.

Supported File Formats

Format Extension Max Size Best For
PDF .pdf 50 MB Manuals, policies, reports
Word .docx 50 MB Training materials, guides
Text .txt 50 MB FAQs, plain content
Markdown .md 50 MB Structured documentation
CSV .csv 50 MB Product catalogs, data tables

Limits:

  • Maximum 100 files per knowledge base
  • Maximum 50 MB per file
  • Processing time: seconds to minutes depending on file size

Step-by-Step Upload Guide

1. Navigate to Your Knowledge Base

  1. Go to Knowledge Base in the sidebar
  2. Click on your knowledge base (or create one first)

2. Open the Upload Section

The "Add Documents" section is at the top of the knowledge base detail page.

3. Select Files

Drag and drop files into the upload area, or click to browse your computer.

You can select multiple files at once.

4. Monitor Processing

Each file goes through these stages:

Stage Duration What Happens
Uploading Seconds File transferred to server
Processing Seconds-Minutes Text extracted, chunked
Embedding Seconds Vectors generated
Indexing Seconds Stored for retrieval

5. Verify Success

Check the document list for status:

  • Completed (green checkmark): Ready for use
  • Failed (red X): See Troubleshooting

File Preparation Best Practices

PDF Files

Do:

  • Use text-based PDFs (not scanned images)
  • Include clear headings and structure
  • Use standard fonts
  • Keep under 50 pages per file

Don't:

  • Upload password-protected PDFs
  • Use PDFs with heavy graphics/images
  • Upload scanned documents without OCR

Word Documents (DOCX)

Do:

  • Use heading styles (H1, H2, H3)
  • Include a table of contents for long docs
  • Use standard formatting

Don't:

  • Embed complex charts
  • Use text boxes extensively
  • Include tracked changes

Text Files

Best for simple, structured content:

# FAQ - Return Policy

## Can I return my order?
Yes, you can return items within 30 days of purchase.

## How do I start a return?
Contact support or use the returns portal.

## When will I get my refund?
Refunds are processed within 5-7 business days.

CSV Files

Great for product catalogs and structured data:

product_name,price,description,features
Widget Pro,$99,Professional widget,Water resistant, 2-year warranty
Widget Basic,$49,Entry-level widget,1-year warranty
Widget Enterprise,$299,Enterprise solution,API access, Priority support

Content Organization Tips

One Topic Per Document

Split large documents by topic for better retrieval:

Instead of Use
company-handbook.pdf (200 pages) leave-policy.pdf, expense-policy.pdf, it-guidelines.pdf
all-products.pdf product-a-specs.pdf, product-b-specs.pdf

Clear Headings

Use hierarchical headings:

# Product Name

## Overview
Brief description of the product...

## Specifications
- Weight: 500g
- Dimensions: 10x10x5 cm

## FAQ

### How do I install it?
Follow these steps...

### What's the warranty?
2 years limited warranty...

Descriptive File Names

Bad Good
doc1.pdf return-policy-2024.pdf
info.docx product-installation-guide.docx
faq.txt customer-support-faq.txt

Updating Documents

To update a document's content:

  1. Delete the old version from the knowledge base
  2. Upload the new version

Note: Document IDs change when re-uploaded. This is expected behavior.

When to Update

  • Policy changes
  • Product updates
  • Price changes
  • New FAQ entries

Keeping Content Fresh

For frequently changing content, consider using Web Crawling instead. Web content can be re-crawled periodically.

Bulk Upload

Upload multiple files at once:

  1. Select all files in your file browser (Ctrl+A or Cmd+A)
  2. Drag them into the upload area
  3. All files will be queued for processing

Tip: Organize files in a folder before uploading to ensure you have the right content.

Processing Status

After upload, each document shows a status:

Status Meaning Action
Pending Queued Wait for processing
Processing Being processed Wait (can take minutes for large files)
Completed Ready Document is searchable
Failed Error occurred Click "Retry" or check Troubleshooting
Stuck Timed out Click "Retry"

Common Issues

"File too large"

Split the document into smaller files (under 50 MB each).

"Unsupported format"

Convert to a supported format:

  • Images → PDF with OCR
  • Old Word (.doc) → DOCX
  • PowerPoint → PDF

Processing takes too long

Large files can take several minutes. If stuck for more than 10 minutes, the system will mark it as "Stuck" and you can retry.

See Troubleshooting for more solutions.

What Happens After Upload?

Once processed, your document is:

  1. Chunked into smaller pieces (~500 tokens each)
  2. Embedded as vectors for semantic search
  3. Indexed in Pinecone for fast retrieval
  4. Available for your agent to search during calls

Learn more: Document Processing

Next Steps

Need help? Check our Troubleshooting guide or contact support.