Enabling Web Browsing for an Agent
To enable web actions for your elvex agent, follow these steps:
Open your agent's configuration settings
Navigate to the Actions section
Locate the Web Browsing option and toggle it on
Save your agent configuration by clicking Save & Publish
Note that while web actions are available to all users, they need to be enabled individually for each agent.
When do Agents use Web Browsing?
Once web actions are enabled, your agent can decide when to perform web searches or read specific URLs based on its internal logic and user interactions. Users can also explicitly ask the agent to search or read a URL if the agent supports that functionality.
To guide your agent's web searching behavior more consistently, you can set up agent rules in the agent's configuration settings. These rules help define when and how the agent should utilize web actions.
The Web Browsing action will work in conversations using the elvex user interface, Slack integration, or API.
When an agent uses the Web Browsing action, you'll see confirmation in the agent's response.
Web Content Retrieval Options
elvex's web browsing actions offer two ways to retrieve content from web pages:
Standard Text Extraction (Default)
By default, when an agent accesses a webpage, elvex extracts the text content and removes all HTML tags, providing clean, readable text. This approach:
Removes formatting, scripts, and styling elements
Focuses on the main textual content
Is ideal for most information retrieval tasks
Works well for reading articles, blog posts, and documentation
Raw HTML Retrieval
When specifically requested, agents can also retrieve the complete raw HTML content of a webpage, including all tags, scripts, and formatting elements. This option:
Preserves the complete HTML structure
Includes all tags, attributes, and embedded content
Is useful for web development tasks, HTML analysis, or when page structure is important
Provides access to metadata, styling information, and embedded scripts
To request raw HTML content, you can ask your agent to "get the raw HTML" or "retrieve the HTML source" of a specific webpage.
Best practices and usage guidelines
When leveraging web searching in your elvex agents, consider the following best practices and guidelines:
Searching the web allows your agent to find up to date information that is outside of the training data in the underlying model or information that is outside of the data in the Datasources that may be connected to that agent.
Provide clear instructions in your agent to explain when and how it should perform web searches or read specific pages.
You may include specific URLs or sites that you want your agent to reference.
Conclusion
Integrating web searching and webpage reading capabilities into your elvex agents opens up a world of possibilities for enhanced functionality and information access. By leveraging web actions, you can create agents that provide users with up-to-date, relevant, and comprehensive information.
As you explore web browsing with your agents, remember to follow best practices and provide rules to guide behavior. If you have any further questions or need assistance, don't hesitate to reach out to the elvex support or consult the documentation.
Happy building!
FAQ
Q: Does Web Browsing work with any AI provider and model?
A: Yes, elvex has designed "actions", like Web Browsing, to work across any AI provider and model your agent is connected with. Actions like Data Analysis are elvex features that we make work on top of the model your agent is connected to. That being said, Web Browsing will work best with more advanced models like ChatGPT-4, ChatGPT-4 Turbo, Claude 3 Opus, Claude 3 Sonnet, and other similarly powerful models.
Q: Can the Web Browsing action access information on web pages that are private or behind a login screen?
A: No, the Web Browsing action can only access information on web pages that are not gated.