What’s a knowledge base?

A knowledge base is simply the place where you store all the information your AI agent can use to answer questions. It can be FAQs, files, or links (URLs) to pages with relevant content.

The AI uses everything in the database to understand context, find the right answers, and give smarter, more accurate replies to users.

You could think of the knowledge base as the AI’s memory—the better and more up-to-date the information you add, the smarter it gets.

This guide shows you how to create a new knowledge base and fill it with content that helps your AI agents answer quickly, correctly, and in the right way.

Create and activate a knowledge base

1. Open Organization settings

   Click the gear icon and choose Organization settings.

2. Go to AI and open Knowledge Bases

   Click “AI” and then Knowledge Bases.

   You’ll see all existing databases that AI agents can use.

3. Create a new database

   Click “+ Create knowledge base”.

   Give it a clear name, for example “Product Info AI” or “Support FAQ AI”.

4. Add FAQs

   Click “Add FAQ” and enter the question and the answer in their fields.

   FAQs help the AI quickly answer common questions.

5. Upload files

   Click “Add file”. You can upload PDF, DOCX, TXT, MD, or CSV.

   These files can include detailed information the AI can read and index.

6. Add web addresses (URLs)

   Click “Add URL” and paste relevant links—like internal guides, external articles, or online documents.

   The AI reads these pages and uses them to build its knowledge.

7. Add Zendesk Help Center (optional)

   If your Help Center is in Zendesk, enter its Help Center URL. The entire Help Center will be indexed in the app so the AI can use the content as a source.

8. Create the database

   Click “Create” to save and activate the database.

   It’s now available for your AI agents to use.

Tips: How to build an effective knowledge base

Once you’ve created your first knowledge base, the next step is to fill it with the right content and structure it so the AI can easily find relevant info. Here are some recommendations:

1. Build databases by context

Create separate knowledge bases for different areas or departments.

That way the AI can answer more precisely, and it’s easier for you to keep each set of information up to date.

Examples:

• Support – troubleshooting guides, FAQs, instructions.
• Sales – product descriptions, pricing, value props.
• Finance – invoice handling, payment terms, contract questions.

This prevents a sales question from getting mixed up with a support answer.

2. Start with well-documented files

Build the foundation with files that explain your processes, routines, and products.

A well-written PDF, DOCX, or TXT file is great source material for the AI to ingest and understand the big picture.

Tips:

• Split longer documents into clear headings.
• Use plain language—AI interprets clear sentences best.
• Update the files regularly.

3. Add FAQs for quick, simple answers

FAQs are perfect for short questions that come up often.

Write each question and answer clearly and focus on the essentials.

Example:

Question: How often is the system updated?

Answer: The system updates automatically every night at 03:00.

Feel free to create multiple FAQs per topic—it helps the AI answer faster.

4. Use URLs for public pages

If you already have guides or articles published online, add their URLs.

The AI ingests that content and uses it as a source when answering questions.

Good examples of URLs:

• Help Center articles.
• Product pages.
• Status pages or news feeds.

5. Keep the knowledge base alive

Treat the knowledge base like a living library.

Review it regularly and update it whenever something changes.

Checklist:

• Remove outdated documents.
• Add new FAQs when you see recurring questions.
• Make sure links and URLs still work.

6. Think holistically

The AI is only as good as the material you feed it.

The more structured, clear, and up-to-date your content is, the better the AI’s answers will be.