Introduction

wolkvox Studio incorporates the new Agentic DB component, also known as Agentic Database or Agentic MySQL DB, an artificial intelligence module designed to interact directly with MySQL databases using natural language instructions.

The component allows a flow to query, insert, update, or delete information, and—when authorized by the guidelines—execute more advanced actions, such as creating tables. This way, the designer can automate data operations without building extensive SQL queries or developing an intermediate API for each action.

Its behavior is controlled from the configuration panel through guidelines that establish what the agent should do, what information it can work with, and what its restrictions are. The specific request can be written directly or received from a variable generated by another component in the flow.

The result of each execution is automatically stored in the predefined variable $ResDBAgentic, so it can be used by subsequent components in the Routing Point.

Agentic DB is available in the following types of Routing Points:

• Voice Flow
• Chat Flow
• Interaction Flow
• Agentic Engine

The component is not available in Agent Scripting.

What Can Agentic DB Be Used For?

The component can be incorporated into processes that need to query or modify information automatically. Some examples include:

• Querying the status of a request, order, account, or service.
• Searching for customer information based on their document or phone number.
• Registering leads obtained during a conversation.
• Inserting results generated by a research component.
• Updating contact data or user preferences.
• Changing the status of a case, ticket, or management process.
• Creating audit or tracking records.
• Validating whether a customer meets certain conditions.
• Querying inventories, balances, availability, or history.
• Deleting or purging records when the process rules authorize it.
• Creating tables or structures necessary for a controlled process.
• Delivering database data to another component in the Routing Point.

For example, a flow can first execute the Deep Research component, store its findings in $ResDeepR, and then send that variable to Agentic DB to structure and insert the information into a MySQL table.

Component Configuration

1. In the sidebar, select the "Cognitive" tab. Within the Agents section, locate the Agentic DB component.
2. Drag it to the design canvas.
3. Double-click the component to open its configuration panel on the right side.

In the "Name" tab, enter an identifier that allows you to easily recognize the component's function within the flow.

Some examples include:

• Query Customer
• Register Prospect
• Update Request Status
• Save Research Result
• Validate Balance

Select the "Description" tab and enter additional information about the component's objective.

You can indicate which database it uses, what operation it executes, or at what point in the flow it should be activated.

Complete the connection fields:

• DB User: Name of the authorized user to access the database.
• DB Password: Password corresponding to the configured user.
• Server IP: IP address of the server where the MySQL database is hosted.
• Database Name: Exact name of the database the agent will work with.

Verify that the information is correct before testing the component.

In the "Agentic DB Guidelines" field, write the general context, objective, and rules that the agent must follow when working with the database.

The guidelines can specify:

• The role the agent should assume.
• The table it can operate on.
• The fields it should query or modify.
• The expected data format.
• Allowed operations.
• Actions it should not execute.
• How to handle missing data.
• Conditions required to insert or update information.

Example:

# Role and Objective
You are an agent specialized in data extraction and database administration.

You must analyze the received information, identify prospect data, and insert it into the `prospects` table.

# Fields
- company
- city
- phone
- email
- address
- notes
- source

# Rules
- Do not invent information.
- When data is not available, use `null`.
- Do not delete records.
- Do not modify tables other than `prospects`.
- Before inserting, validate that the record is not duplicated.

The guidelines must clearly establish the agent's scope. The component will use these instructions as a reference to interpret each request.

In the "DB Request" field, write the specific task the agent must execute.

The request can contain direct text:

Query the customer with document 123456789 and return their name, status, and balance.

It can also use a flow variable:

$ResDeepR

In this case, Agentic DB receives the content stored in the variable and processes it according to the configured guidelines. This allows connecting the component with results from Deep Research, customer messages, forms, previous queries, or other components.

Click "Test Agent" to validate the configuration before incorporating it into the productive flow.

The system generates a local HTML file at a path similar to:

C:/ipdialbox/download/wolkvox_agentic_mysql_test_20260724_132420.html

The test report includes information such as:

• Date and time of generation.
• Component identifier.
• Session memory status.
• Configured server or host.
• Database user.
• Database name.
• Guidelines sent to the agent.
• Processed request.
• Execution result.

The Result field allows you to check whether the operation was successful or why it could not be completed. For example, it may indicate that mandatory data was missing, the request did not contain sufficient information, or the requested operation could not be executed in the database.

Enable the "Enable Session Memory" checkbox when you need the component to retain the context of previous executions during the same session.

• Enabled: The agent can consider previous requests and results.
• Disabled: Each execution is processed independently.

This option can be used in consecutive processes, such as first querying a record and then updating it using the already obtained context.

The "Agentic DB Result Variable" field displays the predefined variable:

$ResDBAgentic

The name of this variable is fixed and cannot be modified.

It stores the response generated by Agentic DB, which may include:

• The records found.
• Confirmation of an insertion.
• The result of an update.
• The number of affected records.
• A validation message.
• An explanation of an error or a non-executed operation.

Subsequently, $ResDBAgentic can be used to display a response to the customer, evaluate a condition, feed another agent, save an additional record, or continue the automation.

Click "Save" to apply the changes made.