Data Sources & Training | Website

Introduction

Website training allows your AI agent to learn directly from your website content.

By adding your website as a data source, you help the agent answer questions using your own pages—such as product information, FAQs, policies, help articles, or documentation—rather than generic responses.

This is one of the most common and powerful ways to train an AI agent in Exei.

What Website Training Does

When you train an agent using a website:

• Exei crawls the provided URLs

• The content is converted into structured knowledge

• The agent uses this information to answer user questions

Website training ensures the agent:

• Gives accurate, brand-specific answers

• Stays aligned with what’s published on your site

• Reduces the need for manual text entry

Where to Add Website as a Data Source

To train your agent using website content:

• Open your AI agent from the My Agents page

• Go to Knowledge Base

• Select Website

This is where you manage all website-based training.

Adding a Website for Training

In the Website section:

• Enter the full URL you want to train the agent on

  • Example: https://yourwebsite.com

• Click Train

Exei will begin fetching and processing the content from the provided URL.

You can add:

• A full website

• A specific page

• Multiple URLs

What Pages Should You Train On?

For best results, start with pages that users commonly ask about.

Recommended pages include:

• Home page (for overall context)

• Product or service pages

• Pricing pages

• FAQs

• Help or support pages

• Shipping, returns, and policy pages

• Terms and conditions

You don’t need to train everything at once—start small and expand as needed.

How Website Crawling Works

When a website is added:

• Exei scans publicly accessible pages

• Content is extracted and structured

The agent does not access:

• Password-protected pages

• Admin dashboards

• Private or gated content

Only content visible to regular website visitors is used.

Training Status and Updates

Once training starts:

• The website shows as processing

• After completion, it becomes active

• If your website content changes:

• You can retrain the same URL

• Updated information replaces older content

Regular retraining is recommended for frequently updated websites.

Using Website Training with Other Data Sources

Website training works best when combined with:

• Document training (PDFs, manuals, policies)

• Text training (specific instructions or exceptions)

• Catalog training (for Shopping Assistants)

Each data source adds more context and improves accuracy.

Best Practices for Website Training

• Start with your most important pages

• Avoid training unnecessary or outdated pages

• Retrain after major website updates

• Use clear, well-structured website content

• Pair website training with AI Persona settings

Good website content leads to better AI responses.

Common Mistakes to Avoid

• Adding too many URLs at once without testing

• Training on incomplete or outdated pages

• Forgetting to retrain after content changes

• Expecting the agent to answer questions not covered on the site

If the content isn’t on your website, the agent won’t know it unless trained elsewhere.

How to Validate Website Training

After training is complete:

1. Open Test AI Agent

2. Ask questions based on your website content

3. Check accuracy and tone

4. Use Instant Retrain if responses need correction

Testing ensures your data is being used correctly.

When to Update Website Training

You should retrain your website when:

• Product or service information changes

• Policies or pricing are updated

• New pages are added

• Users report outdated answers

Keeping training up to date keeps responses reliable.

What’s Next?

Now that your agent is trained using website content, you can expand training further or move to deployment.

Next User Guides:

Data Sources & Training | Documents

Data Sources & Training | Text