Skip to main content

​
Knowledge-Base Overview

A knowledge base (KB) is a structured collection of information that your AI agents can access to answer questions, reference data, or guide conversations during calls. Integrating a knowledge base ensures your agents are informed and provide accurate responses to leads or customers.

​
1. Purpose of Knowledge Base

  • Store FAQs, product/service details, scripts, and client-specific information.
  • Allow AI agents to dynamically reference data during conversations.
  • Reduce errors and improve the relevance and quality of agent responses.
  • Enable consistent messaging across campaigns and agents.

​
2. Types of Knowledge Bases

  • Internal KB: Stored within your platform, typically editable via a dashboard.
  • External KB: Connected via APIs, webhooks, or CRM systems to fetch live data.
  • Hybrid KB: Combines internal and external sources for comprehensive coverage.

​
3. Setting Up a Knowledge Base

  1. Navigate to Knowledge-Base → Create New KB.
  2. Provide a name and description for the knowledge base.
  3. Choose the source type (internal, external, hybrid).
  4. Upload or input data:
    • Text files, spreadsheets, JSON, or structured content.
    • Include FAQs, call scripts, product details, or customer info.
  5. Configure access rules: define which agents or campaigns can use this KB.

​
4. Connecting Knowledge Base to Agents

  • Assign your KB to one or multiple AI agents in the Agent Settings → Knowledge Base section.
  • Agents will automatically reference the KB when generating responses during calls.
  • Set priority or fallback rules if multiple KBs are connected (e.g., internal KB first, then external API).

​
5. Testing & Updating KB

  • Use the playground/test mode to ensure the agent retrieves correct data from the KB.
  • Update the KB regularly to include new information, correct errors, or improve coverage.
  • Keep track of KB versions for rollback or agent re-training if needed.
A well-maintained knowledge base improves agent performance, reduces failed calls, and ensures consistent messaging across campaigns. Start with a small, high-priority dataset and expand over time.