Today, AI Agents have become one of the most powerful tools in the AI ecosystem. They allow you to delegate complex tasks to autonomous systems capable of interpreting, deciding, and executing actions on your behalf. For users, mastering agents means gaining flexibility, autonomy, and far greater capability to build advanced workflows without relying on traditional programming.

In this context, Narrativa® Navigator introduces AI Agents designed specifically for extracting information from the web, enabling you to turn any public webpage into a structured dataset through a simple configuration process. This functionality provides a level of operational freedom that previously required code or external integrations.

Narrativa note: When we refer to AI Agents in Narrativa® Navigator, we are talking about operational entities designed to perform tasks according to your instructions. They do not act independently outside the secure environment of the platform. Their “behavior” is an operational narrative built to help you delegate complex processes in an easy and safe way.

1. What can you do with AI Agents?

Before getting started, it’s helpful to understand why this feature is so powerful:

✔ Total independence

No need for APIs, scripts, or complex technical setups. The agent navigates the web for you.

✔ Customized data extraction

You decide exactly which data points you want and how they should be stored.

✔ Full flexibility

Works with virtually any public webpage.

✔ Time-saving automation

Automates repetitive tasks that you previously had to do manually.

2. How to create a web scraping AI Agent

Below is a step-by-step guide on how to use this feature.

Step 1: Access the Data tab

From the platform sidebar:

1. Click Data.
2. Select Create new Data.

Step 2: Choose the data source type

In the creation wizard:

1. Select AI Agent (Web Scraping).
2. Enter one or more URLs from which you want to extract information.

Step 3: Configure the information the agent should extract

This is where you define how your dataset will be structured.

Add one column for each data point

For every column, specify:

• Column name
• Detailed description, including:
  • which value the agent should look for
  • where it is located on the webpage
  • any useful contextual clues (HTML structure, tags, attributes…)
• Data type such as:
  • Text (default)
  • Numeric
  • Other formats as needed

The agent will analyze the webpage according to your instructions to locate the correct value.

Tip: The clearer and more precise your description is, the more accurate the extraction will be.

Step 4: Preview the extraction

Click Connect to generate a preview of the extracted data.

You will be able to validate:

• Whether the agent is retrieving the correct information
• Whether the format matches what you need
• Whether descriptions need refinement or more columns should be added

You can adjust the configuration at any time. It’s not a rigid or linear process.

Step 5: Configure the dataset as usual

Once the extraction looks correct:

• Select the columns you want to use
• Define the Unique ID
• Set up the Scheduler, same as with any other data source

Final result

You’ll obtain a dynamic dataset, fully tailored to your needs and capable of updating automatically. This dataset can then be used in:

• Workflows
• Narratives
• Content projects
• Analytical processes within Narrativa® Navigator