Skip to main content

Overview

Every AI Agent starts with general knowledge. This is the data used when the LLM was trained. But what if you want it to deeply understand your company's products, services, internal documents, or industry-specific resources?

That’s where Knowledge Bases come in. They allow you to give your AI Agent access to your own content — making it smarter, more accurate, and more aligned with your goals.

What Are Knowledge Bases?​

A Knowledge Base is a collection of documents and content that your AI Agent can access to answer questions more accurately. You can think of it as your Agent’s custom library — full of the information you want it to reference.

Knowledge Bases provide essential context that transforms your Agent from generic to domain-specific.

Supported content types include:

  • PDFs
  • DOCX files
  • Websites (static pages or full domains)
  • YouTube videos (audio and captions are extracted)
  • Video and audio files
  • Other common text-based files

Once uploaded, your content is processed and stored in a way that makes it easy for the AI to search, understand, and use in conversations.

How Knowledge Bases Work​

When you upload documents or link URLs to a Knowledge Base, they go through a data ingestion pipeline behind the scenes. Here’s what happens:

  1. Chunking: The content is broken down into smaller, manageable pieces (chunks) to preserve context.
  2. Vectorizing: Each chunk is transformed into a mathematical format that can be stored in a vector database. This is how the AI “remembers” the content.
  3. Storage: These chunks are stored in your Knowledge Base, ready to be searched whenever a user asks a related question.
  4. Retrieval: When your Agent receives a question, it searches all the Knowledge Bases it’s connected to, finds the most relevant content, and uses it to generate an accurate response.

This process is known as Retrieval-Augmented Generation (RAG) — a powerful way to bring your own data into the conversation.

Uploading Content to a Knowledge Base​

To create a Knowledge Base and upload content:

  1. Go to the Knowledge Bases section in your dashboard.
  2. Click Create Knowledge Base.
  3. Name your KB (e.g. “Product Manuals” or “Customer Support FAQs”).
  4. Upload files or paste links to websites or YouTube videos.
  5. The system will automatically process and store the content — no technical setup required.

You can upload multiple documents into a single Knowledge Base, or keep them in separate ones depending on how you want to organize your content.

Example: You might want one Knowledge Base called “HR Docs” and another called “Product Help Center,” instead of mixing all files into one.

Assigning Knowledge Bases to AI Agents​

Creating a Knowledge Base isn’t enough. For your Agent to actually use the content, you need to assign the Knowledge Base to it.

To do that:

  1. Create or edit an AI Agent.
  2. Go to the Train tab.
  3. Click the Add Knowledge Base button.
  4. In the modal that opens, choose Add Existing Knowledge Base.
  5. Select one or more KBs you want the Agent to use.
  6. Click Save.

From now on, your Agent will search those KBs when answering questions. If no KBs are linked, it won’t be able to reference your documents.

info

You can assign:

  • One Knowledge Base to multiple Agents
  • Multiple Knowledge Bases to a single Agent

This gives you complete flexibility over how your content is used across Agents.

Understanding the Knowledge Base–Agent Relationship​

When an Agent is connected to several Knowledge Bases, it doesn’t favor one over the other. Instead, it:

  • Searches across all linked KBs in parallel
  • Ranks the results by relevance
  • Uses only the most relevant chunks to answer the question

There’s no hard limit to how many KBs you can assign to an Agent, but for performance and clarity, it’s best to only link relevant ones.

If you create an AI Agent without selecting an existing Knowledge Base, the system will automatically generate an empty KB for you. This is a convenience feature in case you prefer to upload documents directly while configuring your Agent.

Organizing Your Knowledge Effectively​

A common question is whether to put all documents into one Knowledge Base or separate them.

Here’s a practical guideline:

  • Group related documents into the same KB
  • Use separate KBs for unrelated topics

This keeps your AI accurate and efficient. For example, combining HR policies with software manuals in the same KB could confuse your Agent when answering questions.

Another key point: if you assign multiple Knowledge Bases to an Agent, it searches them all. This can be helpful, but too many KBs may impact performance slightly — especially if they cover unrelated domains.

Testing Multiple Knowledge Base Versions​

If you're experimenting with different document sets, the best approach is to create multiple AI Agents, each linked to a different set of KBs.

This allows you to:

  • Compare how different KB combinations affect responses
  • Test updated versions of the same KB
  • Control exactly which content is used in each test

Example: You want to test whether a newly revised product guide improves answer quality. Create one Agent linked to the old guide, and another linked to the new one. Ask them both the same question and compare the results.

Best Practices​

  • Always name your Knowledge Bases clearly to keep things organized.
  • Upload documents in formats that preserve structure and text.
  • Keep your KBs updated — re-upload files if the content changes.
  • Assign only relevant KBs to each Agent for best performance.

Summary​

Knowledge Bases are how your AI Agent gets smarter with your content. From manuals to help docs to policies, they allow your Agent to provide answers grounded in your own data.

Just remember:

  • Create a KB, upload your content
  • Link the KB to your AI Agent
  • Ask questions in the chat tab
  • The AI will search and use your documents to give accurate, tailored responses

With the right setup, your AI Agent becomes helpful and fully aligned with your unique business knowledge.