Web Search and Scraping

FormWise includes two built-in web capabilities that let your agents access live information from the internet: web search and web scraping. These are especially useful when your agent needs current, real-time data that goes beyond the AI model's training knowledge.

Both are built in - there's no separate API key to configure. Enabling either one on an Agent node bills a small credit cost per call, the same way other built-in actions do.

Web Search

Web Search lets your agent search the internet for up-to-date information, just like you would in a search engine. It's powered by the Serp API (Google search results).

When to use it

• Your agent needs to answer questions about current events or recent news.
• You want fact-checking against live sources.
• Your agent needs to find information that changes frequently (prices, availability, statistics).

How to enable it

1. Open your Flow in the workflow builder.
2. Click on an Agent node.
3. In the node settings, toggle Web Search on.
4. Save your changes.

Once enabled, the AI decides when to search based on the user's question. You do not need to tell it when to search - if the agent determines it needs current information, it will automatically run a web search and use the results in its response.

Web Scraping

Web Scraping lets your agent read and extract content from specific web pages. It's powered by the Jina Reader API.

When to use it

• Your agent needs to read the content of a specific article or web page.
• You want to extract data from a known URL.
• Your agent needs to research information from particular websites.

How to enable it

1. Open your Flow in the workflow builder.
2. Click on an Agent node.
3. In the node settings, toggle Web Scrape on.
4. Save your changes.

Once enabled, the AI can fetch and read any publicly accessible web page. If a user provides a URL or the agent identifies a relevant page, it will scrape the content and use it to inform its response.

Using both together

Web search and web scraping work great as a pair. For example:

• The agent searches the web to find relevant pages on a topic.
• Then it scrapes the most promising results to read the full content.

You can enable both on the same Agent node for maximum flexibility.

Tips

• Only enable web tools on agents that genuinely need internet access. Not every agent requires live data.
• Each web search or scrape call has a small credit cost - worth knowing if you're enabling it broadly across many agents.
• Web search works best when the user's question involves recent or time-sensitive information.
• Web scraping works with publicly accessible pages only - it cannot access content behind logins or paywalls.