Steps to Create a Knowledge Base

1. Accessing the Knowledge Base Section

• Navigate to the Knowledge Base tab located on the left-hand menu (above the Assistants tab).
• If no knowledge base exists, create one by clicking Create Knowledge Base in the top-right corner.
• Name the knowledge base (e.g., "Health & Fitness Knowledge Base") and confirm.

2. Understanding the Purpose of the Knowledge Base

The AI is connected to OpenAI’s model, which already has general knowledge about industries.

The knowledge base focuses on company-specific information such as:

• Services offered.
• Pricing or estimates.
• Service areas.
• Company-specific FAQs.

3. Adding Data to the Knowledge Base

To populate the knowledge base, you can add data through various methods:

a. Text Input

• Go to Add Data Source > Text.
• Type in details manually, such as: Example: "Our health coaching programs include nutrition, training, accountability, supplementation and ongoing community with support from an expert team."
• Save the entry with a relevant name (e.g., "Our Programs").

b. FAQs

Click Add Data Source > FAQs

.Write common questions and answers. Example:

• Q: "Do you offer free consultation?"
• A: "Yes, we offer free consultation to anyone who's looking to improve their health and wellbeing and is ready to invest in bettering their fitness."
• Q: "What are your services?"
• A: "We specialize in body transformation programs through nutrition, fitness and accountability."

Add as many FAQs as possible - more data improves the AI’s accuracy.

c. File Uploads

• Upload PDFs, images, audio files, or other formats containing company information.
• Example: A PDF with detailed service descriptions.

d. URLs

• Provide a company website URL or a specific webpage.
• The system will extract key information from the page.

4. Testing the Knowledge Base

Use the Test Knowledge Base feature to ensure the AI understands the added data.

Example Query: "What does your program include?"

• AI Response: "Your program includes a year's worth of access, with a weekly plan that covers nutrition, training, and accountability. It's organized into monthly cycles with tasks, weekly reports, and daily checklists."

The system provides a match percentage. Low matches indicate the AI may generate an inaccurate response. Always aim above 55% - ideally 60%+.

5. Refining the Knowledge Base

For incomplete or incorrect responses, copy the query and manually add it to the knowledge base:

• Format the question and answer as an FAQ.
• Save the entry and wait for the system to embed it.

Retest with similar queries to ensure accuracy.

Key Tips for Refinement

• Add as much specific data as possible (FAQs, service descriptions, pricing details, etc.).
• Regularly test and refine responses to improve match accuracy.
• Use deterministic testing to ensure the AI delivers consistent results.