Skip to Content
Knowledge BaseSupported Sources

Supported Sources

Learn about the different content sources you can add to your knowledge base.

Document Types

Upload files directly to your knowledge base.

Supported Formats

FormatExtensionDescription
PDF.pdfPDF documents with text extraction
Word.docxMicrosoft Word documents
Excel.xlsxExcel spreadsheets
CSV.csvComma-separated values
Text.txtPlain text files
Markdown.mdMarkdown documents
JSON.jsonJSON data files

Document Tips

  • Use searchable PDFs when possible (faster processing)
  • Break large documents into smaller files
  • Ensure good text quality for scanned documents
  • Keep content well-structured with headings

Web URLs

Automatically crawl and index web pages.

Adding a Website

Enter your website’s domain URL:

https://example.com

Page Selection

When you add a website:

  1. Enter the domain URL
  2. We discover crawlable pages (up to 10)
  3. Select which pages to include
  4. Optionally add specific page URLs manually

Best Practices

  • Start with your main documentation or FAQ pages
  • Select only relevant pages (skip login pages, admin areas)
  • Re-crawl periodically to capture updates

Data Source Connectors

Sync data from external sources automatically.

Available Connectors

ConnectorDescription
Google SheetsSync rows from Google Spreadsheets
AirtableImport records from Airtable bases
NotionPull content from Notion databases
REST APIFetch data from any REST endpoint
Excel/CSVUpload and parse spreadsheet files

Sync Features

  • Configurable sync intervals
  • Manual sync trigger
  • Sync history tracking

Source Management

Viewing Sources

The Knowledge tab shows all sources with:

  • Source type and name
  • Status (processing, ready, error)
  • Last updated date

Updating Sources

  • URLs: Click “Recrawl” to fetch latest content
  • Files: Upload a new version to replace
  • Connectors: Trigger manual sync or adjust schedule

Deleting Sources

Removing a source immediately removes its content from the knowledge base. This action cannot be undone.

Processing Pipeline

When you add content, it goes through these steps:

  1. Uploading — Content is received
  2. Extracting — Text is extracted from the source
  3. Chunking — Content is split into smaller pieces
  4. Embedding — Chunks are converted to vectors
  5. Indexing — Vectors are stored for search

Processing Status

StatusDescription
PendingWaiting to be processed
ProcessingCurrently being indexed
CompletedSuccessfully processed
FailedProcessing failed (see error message)
Last updated on