Skip to main content

Build AI Knowledge Base

This guide explains how to build and manage the AI Knowledge Base in ConnectNow.

The Knowledge Base enables AI assistants to answer customer questions accurately by accessing structured and unstructured data from multiple sources.

Where You Can Configure the Knowledge Base​

You can configure the Knowledge Base from either of the following locations:

  • During the Setup Wizard (recommended for first-time configuration)
  • From the Dashboard
    • Navigate to Settings
    • Select Knowledge Base

Both options provide access to the same Knowledge Base tools.

Knowledge Base Options in ConnectNow​

ConnectNow supports multiple methods for building the Knowledge Base.
You can enable or disable each option independently based on your requirements.

1. RAG (Retrieval-Augmented Generation)​

RAG allows the AI assistant to answer questions using uploaded documents.

Supported File Formats​

You can upload the following file types:

  • PDF
  • TXT
  • DOCX
  • XLSX
  • CSV

Uploading Documents​

Upload the documents you want the AI assistant to use. The system automatically indexes and analyzes uploaded files, and the AI will query them as needed during conversations.

Tip: Keep documents organized and use clear filenames to make maintenance easier.

2. External API Tool​

The External API tool allows the AI assistant to answer questions using third-party APIs.

Common Use Cases​

  • Weather lookup
  • Order status from external systems
  • Availability checks
  • Any real-time data source exposed via API

You can configure one or more APIs and enable or disable them as needed.

3. MySQL Database Connection​

The MySQL tool allows the AI assistant to answer questions by querying data directly from a MySQL database.

Key Rules​

  • The AI assistant can run read-only queries
  • Insert, update, or delete operations are not allowed
  • The database is used strictly for information retrieval

This option is useful for structured data such as:

  • Product catalogs
  • Pricing tables
  • Inventory information
  • Customer reference data

4. MongoDB Database Connection​

The MongoDB tool enables the AI assistant to retrieve information from a MongoDB database.

Key Rules​

  • The AI assistant performs read-only queries
  • Create, update, or delete operations are not allowed
  • Suitable for document-based data structures

This option is useful for:

  • Dynamic content
  • Log-based data
  • JSON-based records

5. Web Search Tool​

The Web Search tool allows the AI assistant to search the internet and answer questions based on publicly available information.

Typical Use Cases​

  • General knowledge questions
  • Industry-related topics
  • Public information lookups

This tool can be enabled or disabled at any time.

Managing Knowledge Base Tools​

  • Each Knowledge Base tool can be enabled or disabled independently
  • Changes take effect immediately after saving
  • You can combine multiple tools for better coverage

Monitoring Errors and Tool Execution​

ConnectNow provides visibility into Knowledge Base execution issues.

To monitor errors:

  1. Navigate to Settings → Knowledge Base
  2. Review the execution logs and error details

Monitoring helps you:

  • Identify configuration issues
  • Improve data quality
  • Maintain reliable AI responses

Best Practices​

  • Use RAG for organization-specific and static content
  • Use databases for structured and frequently updated data
  • Use external APIs for real-time information
  • Enable web search only when public information is required
  • Keep tool descriptions clear and specific
  • Review error logs regularly

Next Steps​

After building the Knowledge Base:

  • Test AI responses with sample questions
  • Adjust descriptions and configurations if required
  • Proceed with agent testing and go-live steps

Additional guides are available for testing and optimization.