🎯 Guide: Setting Up a User Filter in an ARPIA Session-Based Agent

This guide will walk you through how to configure and test a user-specific filter in an ARPIA agent. The goal is to make sure the agent only shows data that belongs to the user currently logged in.

✅ Step 1: Add a User Email Field to Your Data

First, make sure your data table (or dataset) includes a column with user email addresses. This is how the system knows which records belong to which user.

Example: Add a column named user_email to store each user’s email.

🔧 Step 2: Link the Email Field to the Agent Node

Go to the ARPIA agent and select the node that will query your data. Then, configure it to use the user_email field (or whatever your column is named).

• Make sure this node is selected and ready for setup.

  

▶️ Step 3: Test the Agent Without a Filter

Now, ask the agent a test question without any filters applied. You should see that it returns all records.

• Open AI Governance and check the SQL query the agent created.

• Confirm that there’s no filtering yet — all data is being shown.

  

  Make sure the node is selected on the agent screen:
  

  Ask the agent a question (no filter applied):
  

  In AI Governance, you’ll see the SQL shows all results — there’s no WHERE clause yet:
  

⚙️ Step 4: Add a Filter Based on the User Email

Now apply a filter so the agent only shows results for the logged-in user.

1. Open the node’s settings.
2. Click on "Filtros & Ordenamiento" (Filters & Sorting).
3. Add a filter where user_email equals the session user’s email.

Tip: Use a session variable like {{user.email}} to match the logged-in user.

Note: This approach can also be used with any session user variable, not just emails. For example, you could filter by user_id, user_department, or other session-specific values.

🧪 Step 5: Test Again — Now With the Filter

Ask the same question you used in Step 3. This time, the agent should only return results that match the current user’s email.

• In AI Governance, check that the SQL now includes a WHERE clause using the email.

• This confirms that the session filter is working correctly.

  

🔍 Final Check: Review the Generated SQL

Finally, make sure the SQL logic includes the email filter. This ensures the agent only pulls data for the right user.