Adding data source to your AI agent

Add websites, files, and custom answers to power your AI agent’s responses.

Written By Frieda Yip (Super Administrator)

Updated at August 6th, 2025

To generate accurate and relevant replies, your AI agent must be connected to a knowledge base. This knowledge base acts as the central source of truth The AI agent uses to understand your business, answer customer enquiries, and take appropriate actions.

All AI agents in SleekFlow use the Global knowledge base, which stores all uploaded data sources shared across your workspace. When you add content through an individual AI agent, you’re uploading it to the Global knowledge base — ensuring consistency across agents.

There are 5 options available for adding new sources in the “Configure AI agent” page

  1. File upload
  2. Index website URL
  3. Web search
  4. Custom answers
  5. Select existing data source from Global knowledge base

 

Accessing the “Add new source” modal

You can follow the steps below to access the “Add new source” modal:

  1. In The AgentFlow page, click “Manage” on the AI agent card

     
  2. Click “Knowledge base” on the left-sided menu
  3. You will be redirected to the Knowledge base of The AI agent

     
  4. On this page, you can:
    1. Choose existing data sources from the Global knowledge base
    2. Or click “Add source” to upload new content
  5. A pop-up modal will appear. You can import content by:
    1. Uploading a file (.PDF, .DOCX, .XLSX, or .JPG)
    2. Importing from a website URL
    3. Selecting from existing entries in the Global knowledge base

 

Uploading files to Global knowledge base

Once you have selected “File upload” in the previous step, you can then follow the steps below to upload files to AI agent’s Knowledge base:

  1. In the “Import file to Global knowledge base” page, you can upload files in supported formats such as .DOCX, .PDF, .XLSX, or .JPG. Simply drag and drop your files, or select them from your device. Make sure your files meet the required format and size guidelines.

  2. Once you have reviewed the uploaded files, click “Import”

 

ℹ️ Note:

When you upload a file from within the AI agent’s Knowledge base page, the content is stored in your Global knowledge base. It will be automatically linked to the AI agent you’re configuring, and can be reused across multiple agents in your workspace.


Learn more about managing the Global knowledge base in our Help Center article.

 

 

Indexing website URL as a data source (Recommended for accuracy)

🧠 Choose Index in database for speed and accuracy

Use Website URL indexing when your content is stable — like Help Center articles, policy pages, or product descriptions. The AI will store and reference this content directly, leading to faster replies and higher response reliability.

Ideal for evergreen information that doesn’t change often.

 

 

Use this option to scan and store static website content ahead of time for faster and more accurate replies.

Once you have selected “Website URL indexing” in the previous step, you can then follow the steps below to upload files to AI agent’s Knowledge base:

  1. In the “Index website URL as data source” page, enter the URL that you would like to import to your Global knowledge base. For example: https://sleekflow.io
  2. (Optional) Select how the URL will be scanned:
    1. If you wish to scan all the linked subpages under the URL you entered, you can check the box “Scan all subpages under this URL” under the URL field:
  3. Next, click “Scan URL”. Note that this may take a moment to scan all pages under the URL you have entered
  4. Once the scan is completed, you will be able to expand the toggle to view all pages scanned under the URL
  5. Select the pages you want to import. You can use the checkboxes to include/exclude the pages. You can also search by the title or URL.
  6. Once you have selected the pages that you need to import, you will need to configure how the URL data source be linked to different AI features in SleekFlow. You can link it to:
    1. Existing AI agents in AgentFlow
    2. Inbox co-pilot for message suggestions inside the Inbox
  7. Click “Import” 

Once imported, the content is stored in your Global knowledge base and linked to the selected agent or co-pilot. You can reuse the same data source across multiple agents in your workspace.

 

ℹ️ Note:

When you import a website URL from within the AI agent’s Knowledge base page, the content is stored in your Global knowledge base. It will be automatically linked to the AI agent you’re configuring, and can be reused across multiple agents in your workspace.


Learn more about managing the Global knowledge base in our Help Center article.

 

 

 

Web search as a data source (Recommended for real-time use)

⚡ Choose Web search if your content changes frequently

 

Use Web search when your website is updated regularly — such as blog posts, real-time pricing, or limited-time offers. This ensures your AI agent always pulls the most current version of the page.

Just note: content is fetched live during each conversation, which may slow down response time and reduce accuracy slightly.

 

 

 

Once you have selected “Website search” in the previous step, you can then follow the steps below to import website URL to AI agent’s Knowledge base:

  1. In the “Search website in real time as data source” page, enter the URL that you would like to import to your Global knowledge base. For example: https://sleekflow.io

 

  1. Once you have entered the URL, you will need to configure how the URL data source be linked to different AI features in SleekFlow. You can link it to:
    1. Existing AI agents in AgentFlow
    2. Inbox co-pilot for message suggestions inside the Inbox
  2. Click “Import” 

Once imported, the content is stored in your Global knowledge base and linked to the selected agent or co-pilot. You can reuse the same data source across multiple agents in your workspace.

ℹ️ Note:

When you import a website URL from within the AI agent’s Knowledge base page, the content is stored in your Global knowledge base. It will be automatically linked to the AI agent you’re configuring, and can be reused across multiple agents in your workspace.


Learn more about managing the Global knowledge base in our Help Center article.

 

 

Adding custom answers to Global knowledge base

💡 Custom answers are matched by context, not exact keywords

The AI uses large language model (LLM) capabilities to understand the meaning behind a customer’s message — not just the exact words. As long as the message aligns with the intent of your training question, the AI will detect it and respond with your configured answer.

 

This means you don’t need to list every possible phrasing. Instead, focus on providing clear examples of intent and a strong, reusable answer.

 

 

Once you have selected “Custom answers” in the previous step, you can then follow the steps below to add custom answers to AI agent’s Knowledge base:

  1. You will be redirected to the “Add custom answers as a data source” page
  2. In this page, first fill in the “Data source name”. This name helps you identify the custom answer later and will be shown in your “Global knowledge base” alongside other sources like files or website URLs.
  3. Next, fill in the following fields:
    1. Question
      1. Write a typical customer question that your AI agent should be able to recognize and respond to. This doesn’t need to be an exact phrase — the AI uses contextual understanding to match similar messages during live conversations. Example: “What’s your refund policy?”
      2. You can improve accuracy by adding more variations using the “Add training question button”. These help the AI better understand different ways a customer might ask the same thing.
    2. Answer
      1. Provide the full response you want the AI to deliver. This should be clear, accurate, and helpful — including explanations, steps, or policy details as needed. Example: “We accept refund requests within 7 days of purchase. Please contact our team with your order number to start the process.”
  4. Once you have filled in the answer, click “Add”

 

Selecting data sources from Global knowledge base to an AI agent

Once you have selected “Selected from Global knowledge base” in the previous step, you can then follow the steps below to import website URL to AI agent’s Knowledge base:

  1. In the “Select data sources for this AI agent” page, you will see the existing data sources in your Global knowledge base

  2. Click on the checkbox next to the data source you would like to link this AI agent to
  3. Once you have reviewed the selected data sources, you can click “Select”