Skip to main content

Overview

AI Knowledge is Wegent's knowledge management feature that allows you to create and manage structured knowledge for AI agents to retrieve and reference during conversations.

Knowledge Demo

πŸ“‹ Core Concepts​

Knowledge Base​

A Knowledge Base is a container for storing and managing knowledge:

  • Document Collection: Contains multiple related documents
  • Vector Storage: Document content converted to vectors for semantic retrieval
  • Access Control: Supports private and public access

Knowledge Base Types​

Wegent provides two types of knowledge bases:

TypeDocument LimitChat SupportBest For
Notebook50 documentsβœ… YesSmall knowledge bases, interactive Q&A
ClassicUnlimited❌ NoLarge document libraries, batch retrieval

Retrievers​

A Retriever defines how to search for information in the knowledge base:

  • Semantic Retrieval: Intelligent retrieval based on vector similarity
  • Keyword Retrieval: Traditional keyword matching
  • Hybrid Retrieval: Combines the advantages of both approaches

🎯 Main Features​

1. Knowledge Base Management​

  • Create multiple knowledge bases
  • Choose knowledge base type (Notebook/Classic)
  • Upload and manage documents
  • Automatic document chunking and processing
  • View processing status and statistics

2. Document Support​

Supports multiple document formats and data sources:

SourceDescription
File UploadSupports Markdown, PDF, Word, plain text
Text PastePaste text content directly
External TableImport from DingTalk/Feishu tables
Web ScrapingAutomatically scrape content from URLs

3. Chunking Strategies​

Flexible document chunking options:

StrategyBest For
Smart ChunkingGeneral documents, auto-detect structure
Sentence-basedFAQ, Q&A content, precise matching
SemanticLong articles, maintain topic coherence

4. Retrieval Configuration​

  • Choose retrieval strategy (semantic/keyword/hybrid)
  • Set number of returned results
  • Configure relevance threshold
  • Retrieval test feature

5. Summary Features​

  • Document-level summaries: Auto-generate document summaries
  • Knowledge base-level summaries: Overall knowledge base overview
  • Summary retry: Retry when generation fails

πŸ“– Documentation Navigation​

DocumentDescription
User GuideComplete knowledge base guide
Knowledge Base TypesNotebook vs Classic comparison
Document ManagementAdding and managing documents
Chunking StrategiesDocument chunking strategies
Configuring RetrieversRetrieval strategy and parameter configuration

πŸš€ Quick Start​

Create Your First Knowledge Base​

  1. Navigate to the Knowledge page
  2. Click New Knowledge Base
  3. Choose knowledge base type:
    • Notebook: For small knowledge bases, supports chat
    • Classic: For large document libraries
  4. Fill in name and description
  5. Upload document files
  6. Wait for document processing to complete

Configure Retrievers​

  1. Enter knowledge base details
  2. Click Retrieval Settings
  3. Select retrieval strategy:
    • Semantic: Suitable for conceptual questions
    • Keyword: Suitable for exact matching
    • Hybrid: Balances both approaches
  4. Use Retrieval Test to verify effectiveness
  5. Adjust parameters and save

Use in Agents​

  1. Go to agent settings
  2. Select created knowledge bases in the knowledge base options
  3. Save configuration
  4. Conversations with this agent will automatically reference knowledge base content

πŸ’‘ Use Cases​

Product Documentation​

  • User Manuals: Store product usage instructions
  • FAQ Collections: Common questions and answers (Notebook recommended)
  • Release Notes: Product update logs

Technical Documentation​

  • API Documentation: Interface definitions and usage examples
  • Architecture Docs: System design documentation
  • Best Practices: Team experience summaries

Enterprise Knowledge​

  • Policies: Company internal regulations
  • Training Materials: Employee training documents (Notebook recommended)
  • Project Documentation: Historical project archives (Classic recommended)

πŸ”— Related Resources​