The AI Knowledge Base is the central repository of information that the Assistant uses to generate accurate, relevant, and contextual replies. By structuring and integrating multiple content sources, the Knowledge Base ensures that responses reflect your product data, documentation, and custom content.

Overview

The AI Knowledge Base enables the Assistant to:

Retrieve product-specific information
Reference documentation and supporting materials
Provide context-aware answers
Deliver consistent, up-to-date responses

It combines both general sources and product-related sources to create a unified information layer.

Knowledge Base Structure

The Knowledge Base consists of two main categories:

1. General Sources

General sources provide broad information that the Assistant can use across all products and conversations.

Supported general sources include:

Product catalogue
Arbitrary text documents (plain text)
PDF files
Web URLs

These sources supply foundational knowledge that is not limited to a specific product.

2. Related Sources

Related sources are tied to specific products. These allow the Assistant to provide targeted, product-aware responses.

Examples include:

A PDF manual linked to a specific product
Product-specific documentation
Technical sheets or usage guides assigned to a particular item

When a user asks about a specific product, the Assistant can prioritize its related sources for more accurate and contextual answers.

Integrating Content into the Knowledge Base

Content can be integrated into the AI Knowledge Base in the following ways:

Text Feed

General sources and text-based content can be integrated using the Text Feed.

This method is ideal for:

Information from web pages, like basic information, blogs
FAQs
Arbitrary plain text documentation

Item Feed

When content is related to specific products, it can be integrated through the Item Feed as additional custom properties.

This approach is used for:

Product-specific manuals
Attached documentation (e.g., PDFs)
Supplementary product information
Custom attributes extending standard product data

How the Assistant Uses the Knowledge Base

When generating replies, the Assistant:

Identifies relevant products (if applicable)
Retrieves related product-specific sources
Supplements with general sources when needed
Synthesizes the information into a coherent response

This layered approach ensures:

Accurate product answers
Context-aware explanations
Reduced hallucinations
Consistent information delivery

Best Practices

To maximize the effectiveness of the AI Knowledge Base:

Keep product catalog data accurate and structured
Attach relevant documentation directly to products
Use clear and well-formatted text in general sources
Regularly updating FAQs is an effective way to learn the right answers.
Regularly update PDFs and linked URLs

Well-organized content directly improves response quality.

Customization

Additional sources can be added to the Assistant (additional API adapter, data connector to get Q&A from email, orders API, etc.)