Documentation

Data Management

The Data Management page lets you view, search, test, and manage your chatbot's knowledge base. Access it from Admin > Data in your admin panel.

Overview

Your chatbot's knowledge comes from two data sources:

TypeDescriptionSource
DocumentsText content with AI-searchable embeddingsWebsite crawler, document uploads
ProductsStructured product catalogCSV upload, API sync

Search Modes

RAG Query Search

The most powerful way to understand what your chatbot knows. This simulates exactly what the AI sees when answering customer questions.

How to use:

  1. Enter a question (e.g., "Show me products under $500")
  2. Click Run RAG Query
  3. Review the documents and products the AI would receive

Results include:

  • Search performance metrics (embedding time, search time)
  • Extracted keywords and filters
  • Detected price/budget ranges
  • Matching documents ranked by relevance
  • Matching products based on filters

Use RAG Query to:

  • Verify the right content appears for common questions
  • Debug why certain products aren't showing
  • Test price and category filtering
  • Fine-tune search quality

Database Search

Direct text search across all your data. Useful for finding specific items by URL, title, or category.

Documents Tab

View your knowledge base content organized by source.

View Modes

ModeDescription
By DocumentOne row per source URL - overview of all content
By ChunkOne row per text segment - detailed inspection

Document Information

Each document shows:

  • Title - Page or document title
  • URL - Original source location
  • Chunks - Number of text segments
  • Tokens - Total content size
  • Date - When content was indexed

Managing Documents

Click any document to view details and options:

  • View the full content
  • See all chunks from this source
  • Delete Document - Remove outdated or incorrect content

When to delete documents:

  • Content is outdated or incorrect
  • The source page no longer exists
  • Duplicate content is causing confusion

Products Tab

View and manage your product catalog.

Browsing Products

  • Use category chips to filter by category
  • Each chip shows the product count
  • Click a product row for full details

Product Information

FieldDescription
ProductTitle and link to product page
PriceProduct price
NumericCustom attributes (size, weight, year, etc.)
CategoryProduct category
MediaIndicates attached image/video

Product Details

Click any product to view:

  • Attached media (image, video, or audio)
  • Full product information
  • Search keywords
  • Raw data record

Attaching Media to Products

Products with images get better engagement in chat:

  1. Click a product row to open details
  2. Click Add Media or Replace Media
  3. Select from your media library or upload new
  4. Media displays when the product is recommended

Deleting Products

  1. Click a product row to open details
  2. Click Delete Product
  3. Confirm deletion

RAG Results

After running a RAG Query, additional tabs appear:

RAG Vectors Tab

Shows knowledge base content retrieved for your query:

ColumnDescription
RankRelevance order (1 = most relevant)
DocumentTitle, URL, and content preview
DistanceSimilarity score (lower = more relevant)

Understanding distance scores:

  • 0.0 - 0.3 - Very similar (near-exact match)
  • 0.3 - 0.5 - Similar (related content)
  • 0.5 - 0.80 - Somewhat similar (possibly relevant)
  • Above 0.80 - Low similarity (typically filtered out)

RAG Products Tab

Shows products matching your query's keywords and filters:

  • Ranked by relevance
  • Includes price, category, and attributes
  • Shows which filters were applied

Troubleshooting

"Why doesn't my chatbot know about X?"

  1. Use Database Search to find the content
  2. If not found: Content hasn't been crawled or uploaded yet
  3. If found: Use RAG Query to test if it appears in search results
  4. If not appearing: Content may not be relevant enough to the query

"Products aren't showing for price queries"

  1. Run a RAG Query with a price (e.g., "items under $200")
  2. Check if a price was detected in the results
  3. Verify products in that price range exist in your catalog
  4. Check that products have prices assigned

"Wrong content is appearing"

  1. Use RAG Query to identify the problematic content
  2. Open the document or product details
  3. Delete incorrect or outdated items
  4. Re-crawl or re-upload correct content

"I need to refresh my knowledge base"

  • Website content - Use the Crawler to re-crawl your site
  • Documents - Re-upload via Data Uploads
  • Products - Re-upload your product CSV

Best Practices

  1. Test regularly - Use RAG Query to verify answers to common questions
  2. Clean up content - Remove outdated information that might confuse the AI
  3. Add product media - Products with images perform better
  4. Check categories - Ensure products have correct categories for filtering
  5. Monitor quality - Review RAG results periodically to maintain accuracy

Related Pages