AI Hub – Knowledge Base

A Knowledge Base in ihakimi is a centralized repository of information that enables AI agents and automation flows to access your business-specific data to provide more accurate and contextualized responses to users.

Knowledge Base Structure

A Knowledge Base consists of two main components:

1. Vector Store: The main container that groups and organizes multiple related files
2. Files: Individual documents containing your business information (PDFs, text documents, etc.)

Important: Knowledge Bases are shared among all bots in the workspace, enabling efficient information reuse.

How to Create a Knowledge Base

Prerequisites

• Connect your OpenAI API Key at: 

Workspace → Integrations → Artificial Intelligence → OpenAI

Step-by-Step Guide

1. Access AI Hub
   • Enter any chatbot
   • Navigate to AI Hub → Knowledge Base

2. Create a Vector Store
   • Click “Create New Vector Store”

Setting Up the Vector Store

• Define a name (e.g., “Business Info”)
• Set expiration (optional – 0 = no expiration)
• Upload files directly or select existing files

3. Add Files
   • Click “Files” to add more files

4. Create the Knowledge Base
   • Click the “+” button

• Name your Knowledge Base
• Add description (optional)
• Select the created Vector Store
• Save settings

Where to Use Knowledge Base

1. AI Agents

Location: AI Hub → AI Agents 

→ Model Responses → OpenAI Responses

Configuration:

• Enable “Knowledge Base” option
• Select desired Knowledge Base from dropdown
• AI Agent will automatically query these files to answer user question

How it Works: Whenever the AI Agent needs references to create responses about your business, it will automatically access the files in the selected Knowledge Base.

2. Flow Builder

Location: Flow Builder → Actions → Integrations → OpenAI

→ Search Knowledge Base

Available Settings:

• Model Response: Select the response model
• System Message: Configure AI prompt (how AI should respond based on business information)
• User Input: Usually uses the last user text input
• Knowledge Base: Select the knowledge base
• Max Number of Results: Number of results returned (default: 2)
• Max Tokens: Minimum 1,000 tokens recommended for complete responses
• Remove Key Values: Option to remove specific parameters from payload (useful to stay within 20,000 character limit for JSON fields)

Advantages of Creating a Knowledge Base

1. Accurate and Consistent Responses

• AI provides exact information about your business, products, and services
• Eliminates generic or inaccurate responses
• Maintains consistency in shared information

2. Centralized Management

• Update information in one place
• Changes automatically reflect across all workspace bots
• Simplifies maintenance and version control

3. Time Savings

• Eliminates need to train each bot individually
• Efficient content reuse across multiple bots
• Reduces setup time for new agents

4. Scalability

• Easily add new files as your business grows
• Organize information in different Vector Stores by category
• Support for multiple knowledge bases for different contexts

5. Enhanced User Experience

• More relevant and detailed responses
• Reduction in unanswered questions
• More efficient and personalized support

6. Integration Flexibility

• Use in AI Agents for automatic responses
• Integrate into complex flows in Flow Builder
• Combine with other actions and integrations

Best Practices

1. File Organization
   • Keep related files in the same Vector Store
   • Use descriptive names for easy identification
   • Regularly update information
2. Token Configuration
   • For Knowledge Base searches, use minimum 1,000 tokens
   • Adjust according to response complexity needs
3. Monitoring
   • Check file processing status
   • Use “Sync” to refresh view after changes
   • Test responses after configuration