👤 For pipe admins
🔐 Available on all plans
🎯 For those who want to empower their AI Agents with real-time web search capabilities to build a more autonomous digital workforce.
Web Search is an advanced tool that allows Pipefy's AI Agents to perform internet searches to collect updated information and context not present within your process.
By enabling this function, the agent gains the "skill" of consulting external sources to support its decisions and fill out fields.
What this tool solves
By default, an agent's performance is restricted to the information in your pipe and attached documents. While this limit is essential for maintaining data control and security (ensuring solid governance), Web Search allows you to expand this frontier when external context is needed.
Bring in external context: Allows the agent to validate information or search for complementary data on the web in real-time.
Eliminate information silos: Connects your process to trends, quotes, news, and public market data.
Increase accuracy in dynamic tasks: Ideal for tasks that depend on external variables that change constantly.
Step-by-step configuration
The setup is simple and follows the same pattern as other advanced tools (such as Intelligent Document Processing):
In the desired pipe, access your AI Agents
Create a new agent or edit an existing one
In the configuration side panel, click on the Behaviors tab
Locate the Advanced tools section
Switch the Web search toggle to ON
Now, in the Detail the instructions field, instruct the AI on how and when to use this new capability
ℹ️ Tip: The AI will autonomously decide when external search is necessary. To ensure accuracy, be direct: "If the [Supplier] field is empty, use Web Search to find the official website and fill in the URL field." or "Consult recent news about the company [Name] to assess reputational risk."
💡 Get Inspired: Use Cases
Supercharge your processes with global information: perform reputational risk assessments with real-time news, validate supplier prices by comparing quotes with the market, enrich leads with professional LinkedIn profiles, and consult business reviews or hours directly from Google.
Real-world examples:
Risk Analysis: Search for the latest news on banks or partners to identify financial crises or legal issues before closing a contract.
Procurement: Validate if prices received in quotes are aligned with the average found on the Web.
Market Research: Collect customer reviews, operating hours, and traffic patterns for specific establishments.
Behaviors and Costs
Pricing: Using Web Search consumes an additional 0.5 AI credits per successful execution.
Web Search vs. Web Scraping: Web Search (available now) performs general searches to extract summaries and links. The Web Scraping feature (deep access to pages for granular data extraction) is still under development.
How to test and validate
Since this is a dynamic action, the best way to validate it is through real-world execution:
Adjust the Prompt: In an agent, add a clear instruction that depends on external data.
Map the Fields: Ensure there are text fields ready to receive the results (e.g., Search Summary, Source Link).
Trigger the Automation: Move a card or create a new one to trigger the agent.
Verify the result: If the agent does not bring back the information, try refining the prompt to be less ambiguous about the search term.
If you have any questions or would like to share your thoughts on the Web Search experience, click here to fill out our form and help us keep improving our agents.