The Generic Smarton is the most flexible of the three types available in Atom. It adapts to different needs, as you can customize it with prompts and, if you wish, connect it to a knowledge base.

It is ideal for:

In Flowbuilder, drag a Smarton to the canvas of your flow.

In the configuration menu, choose the Generic option within Smarton Type.

The prompt is the "instruction" that will tell the Smarton how to behave.

Here you can:

To configure the prompt, you must access "Smarton Customization", as indicated in the image:

💡Tip: the more detailed and contextualized the prompt, the better the bot's response will be.

A good practice is to complete these sections when creating the prompt:

[Personality]

Define the identity of the bot: name, company name, communication tone, and personality traits you want it to have (formal, close, technical, etc.).

[Task]

Specify the concrete tasks that the bot will be in charge of.

For example: advising the user on which product is the most suitable according to their need.

[CTA]

Indicate how you want the bot to close conversations.

For example: always end by asking if the user wants to proceed with the purchase or talk to an advisor.

[Format]

Define guidelines on the format of the responses, such as:

[Restrictions]

Establish the topics that the bot should not respond to or any other necessary limitation.

[Examples]

Include examples of expected responses to certain questions.

This is especially useful for defining greetings, initial presentations, or the way in which the bot offers the menu of available options.

You can upload and use more than one knowledge base so that the Smarton can answer both specific queries and general company questions (for example: business hours, return policies, payment methods, etc.).

This allows you to separate the information by type and improve the quality of the responses.

To learn more about how to manage your knowledge bases, we recommend learning more about the Resource Manager.

📌Good practice:

Separating the information into different knowledge bases improves the AI's understanding, makes the responses more accurate, and facilitates content maintenance in the future.

Add branches according to what you want to detect in the conversation.

It is important to include the words or expressions that users are likely to use, as they will determine which way the flow will continue.

For example, if you want to detect interest in spare parts, you can configure the intention with terms such as:

"spare parts", "engines", "parts", "windows"

This way, when the user mentions any of these words during the conversation, the flow will automatically go down the corresponding branch.

You can configure what data you want to record during the conversation. It is important to keep in mind that the data will only be captured if the user mentions it explicitly during the interaction.

The objective of this Smarton is not to request information from the user, but to automatically detect and save data that arises naturally in the conversation.

If the field does not exist, you can create it indicating name, description, and data type.

Classify the user according to the conversion funnel:

Example:

Define what happens if the user does not respond.