Skip to main content

What is a Knowledge Base?

A knowledge base is the foundation of your agent’s intelligence. It’s a collection of information sources that your agent draws from when answering visitor questions. The quality and relevance of your knowledge base directly determines how helpful your agent will be.
Think of the knowledge base as your agent’s memory. The more relevant, well-organized content you provide, the better your agent can assist visitors with accurate, contextual answers.

How It Works

When a visitor asks your agent a question, here’s what happens behind the scenes:
1

Query Processing

The visitor’s question is analyzed and converted into a format that can be searched against your knowledge base.
2

Content Retrieval

The system searches through all your knowledge sources to find the most relevant content pieces that match the query.
3

Reranking

Retrieved content is reranked by relevance to ensure the best information is prioritized.
4

Response Generation

The AI generates a natural, conversational response using the retrieved content as context, ensuring accuracy based on your actual content.
This approach is called Retrieval-Augmented Generation (RAG) and ensures your agent provides accurate answers grounded in your content rather than making things up.

Source Types

Your knowledge base can include three types of content sources:
Source TypeDescriptionBest Use Cases
Web SourcesPages imported from your website via URL, sitemap, or crawlProduct pages, documentation, blog posts, service descriptions
Text SourcesCustom content you write directly in the platformInternal knowledge, company policies, specialized information
Q&A SourcesSpecific question-answer pairs you defineFAQs, common support queries, precise answers for specific questions

Managing Sources

All your knowledge sources are managed from the Sources tab in your agent’s dashboard. From there you can:
  • Add new sources of any type
  • View all existing sources and their status
  • Edit source content (for text and Q&A)
  • Delete sources you no longer need
  • Monitor ingestion status and processing
When you add or update sources, your agent’s knowledge base is automatically updated. Changes typically take effect within a few minutes after processing completes.

Source Status

Each source has a status indicating its current state:
StatusMeaning
PendingSource is queued for processing
ProcessingContent is being extracted and indexed
ReadySource is active and available to your agent
FailedProcessing encountered an error (check source details)

Best Practices

Adding hundreds of irrelevant pages can actually hurt your agent’s performance. Be selective and include only content that visitors would actually ask about.
Group related information together. If you have product pages, support articles, and company info, make sure each is well-structured and covers topics comprehensively.
Remove outdated sources and add new content regularly. An agent giving outdated information can frustrate visitors and damage trust.
Web sources are great for existing content, but supplement with Q&A pairs for common questions and text sources for information not on your website.
After adding new sources, test your agent in the Playground with relevant questions to verify it’s using the new content appropriately.

Common Scenarios

Scenario 1: Documentation Website

If you have technical documentation, import your docs pages via sitemap. Supplement with Q&A pairs for the most common “how do I…” questions.

Scenario 2: E-commerce Site

Import product pages and category descriptions. Add Q&A pairs for shipping policies, return processes, and sizing guides that might not be in your main content.

Scenario 3: Service Business

Import your services pages and case studies. Add text sources for internal processes and Q&A pairs for pricing and availability questions.

Next Steps