Knowledge Base

Knowledge Base: How to Train Your Flora AI Chat Agent

Flora AI’s Knowledge Base is where you store information about your business, products, services, and offers so your Chat agent can give accurate, context-aware answers to users.

By adding high‑quality, relevant content to the Knowledge Base, you significantly improve the Chat agent’s ability to understand questions and respond correctly.

What Is the Knowledge Base?

The Knowledge Base is a central repository of information that your AI uses when generating responses. You can add content in several ways:

• Manually pasted text

• Uploaded files

• Scraped website content

• Question & Answer (Q/A) pairs

Each method helps “train” the Chat agent on the information that matters most to your users.

Adding Text Manually

TEXT is content you copy and paste directly into the Knowledge Base.

Use this option when:

• You have content in emails, documents, or notes that you can easily copy.

• You want to add short, focused pieces of information (e.g., product descriptions, policy snippets, pricing details).

Best practices:

• Keep each text entry focused on a single topic.

• Remove irrelevant or outdated information before pasting.

• Use clear headings and bullet points where helpful.

Uploading Files

You can upload text-based documents to quickly add larger amounts of content.

Supported file types:

• .PDF

• .TXT

• .DOC

• .DOCX

Important:
Make sure your PDF contains selectable text, not just an image of text. If you cannot highlight or copy the text in your PDF, the AI will not be able to read it. In that case, convert or OCR the file first.

When to use file uploads:

• To import manuals, guides, policies, or reports.

• To add existing documentation without copying and pasting.

Tips for better results:

• Remove sections that are not relevant to user questions (e.g., legal boilerplate, unrelated appendices).

• Keep documents up to date and re-upload when major changes occur.

Scraping Website Content

You can pull text directly from your website into the Knowledge Base.

How it works:

1. Enter the website URL you want to scrape.

2. Select the maximum number of website URLs (pages) you want to scrape.

3. The system pulls the text content from the page(s) and adds it to the Knowledge Base.

Use cases:

• Importing product pages, service descriptions, or FAQ pages.

• Quickly syncing your AI with your public website content.

Important considerations:

• More data is not always better.
  Many websites contain navigation text, legal footers, blog sidebars, or unrelated content that users will never ask about.

• Adding too much random or irrelevant text can:

• Confuse the AI

• Lead to less accurate or off-topic responses

Best practices:

• Start with your most important pages (e.g., pricing, features, FAQs, support).

• Avoid scraping pages with mostly navigation, ads, or unrelated content.

• Periodically review and clean up scraped content to keep it relevant.

Using Q/A Pairs

Q/A refers to saving and managing knowledge content in a question and answer format.

This is one of the most powerful ways to train the AI on exactly what users might ask and how you want it to respond.

Why Q/A Is So Effective

• Highly targeted: Each entry directly maps a user question to a specific answer.

• Structured: The format is consistent, making it easier for the AI to match user queries.

• Precise responses: Especially effective for common, repeatable questions (e.g., “What are your support hours?”).

Because of this structure, Q/A content often yields more accurate answers for specific queries than unstructured text.

How Q/A Works in Flora AI

• You manually create question and answer pairs for known or expected user questions.

• As users interact with the Training Chat Agent function, new or improved Q/A pairs can be created and updated.

• All updated questions and answers from training sessions are stored in the Q/A section of the Knowledge Base.

When to Prioritize Q/A Format

Q/A is especially useful if:

• You have an FAQ or help center with clearly defined questions.

• You know the exact wording of common user questions.

• You want tight control over how certain answers are phrased.

If your Knowledge Base is largely FAQ-style content, prioritizing Q/A will improve both efficiency and accuracy.

Limitations of Q/A

• Less flexible: Because it’s structured, Q/A is not as adaptable as free-form text or long documents.

• Coverage dependent: It works best when you anticipate the main questions users will ask.

For broader or more exploratory content (e.g., long guides, articles), combine Q/A with text, files, or website content.

Choosing the Right Content Type

Use this quick guide to decide which method to use:

• TEXT (manual paste):

• Short, focused information

• Quick updates or corrections

• Content from emails, notes, or small docs

• FILES (PDF, TXT, DOC, DOCX):

• Long-form documents (manuals, policies, guides)

• Existing documentation you want to import in bulk

• WEBSITE (URL scraping):

• Public-facing content already on your site

• Product pages, service descriptions, FAQs

• Q/A pairs:

• FAQs and common support questions

• Critical information where answer wording matters

• Content refined through Training Chat Agent interactions

Data Quality and Maintenance

Regardless of the method you use, the quality of your Knowledge Base directly affects the quality of your AI’s responses.

Keep in mind:

• Add relevant, clean data only.

• Remove outdated or incorrect information.

• Regularly review Q/A pairs and other content based on real user interactions.

• Update the Knowledge Base whenever your products, services, or policies change.

By thoughtfully curating your Knowledge Base and leveraging text, files, website content, and Q/A pairs, you ensure your Flora AI Chat agent delivers accurate, helpful, and context-aware responses to your users.