Websites

Website Knowledge

This knowledge can be used in the Knowledge Step in the AI Agent builder

Adding Website Knowledge to EasyBeam

The Add Website feature in EasyBeam allows you to add knowledge from a website to your AI agent. This feature includes configurable encoding strategies and an optional web crawler to retrieve structured information efficiently.

Opening the Add Website Modal

To add website knowledge:

  1. Open the Knowledge page (the book on the left nav)

  2. Open the Add Website modal.

  3. Enter the URL of the website you want to process.

  4. Configure the encoding strategy and optional crawling settings.

  5. Click Add Website to submit.


Encoding Strategies

EasyBeam offers three encoding strategies to optimize how website content is chunked and stored for retrieval:

Strategy
Chunk Size
Overlap
Best For

Precise

256

64

Technical documentation, code

Balanced

512

128

General content like articles and blogs

Context-Heavy

1024

256

Long-form narrative content, complex explanations

Each strategy balances chunk size and overlap to ensure effective retrieval when querying the AI.


Website Crawler (Beta)

The Website Crawler allows you to fetch multiple pages from a website by following links within a defined scope.

Enabling the Crawler

  • Toggle the Website Crawler (Beta) switch to enable crawling.

  • Configure additional crawling options.

Crawl Options

Option
Description

Crawl Depth

Limits how many levels of links are followed (currently 1-2).

Max Pages to Crawl

Defines the number of pages the crawler will retrieve (min: 10, max: 500).

Use JavaScript

Enables JavaScript rendering for dynamically loaded content.

Crawl Scope

Defines which links the crawler will follow.

Crawl Scope Options

Scope
Description

Same Domain

Only follows links on the same domain (e.g., example.com).

Same Root Domain

Follows links across subdomains (e.g., *.example.com).

All Links

Follows any external link (use with caution).


Advanced Options

URL Pattern Matching

You can specify a URL pattern (using a regular expression) to limit the pages the crawler follows. Example:

Last updated