Chatbot - Using the Copilot Token

How to use the {copilot=...} token to configure smart AI responses based on prompts and dynamic fields.

To use this feature, you need to have a configured key and a selected model in "System Settings" > "Artificial Intelligence APIs" > "Assistant Model".

Saving the lead's response

In this example, I created a scenario in which, when the lead informs the number of days they intend to rent a car, they will receive the total rental cost.

For this, you can start by asking the lead how many days they want to rent the vehicle. In the question block, configure it to save the answer in the variable number_of_days.

Send Question block.

Configuring the response message

By using the token {copilot=...}, we can insert line breaks within the content to make the prompt more organized and readable. It is worth noting that the token can be used anywhere in the "Send Message" block, according to your preference.

In addition, it is possible to define the level of creativity of the AI's response directly in the token, using the prefixes:

• temp__creative – for more creative responses with greater variation (more "random").

• temp__deterministic – for more direct and predictable responses (more "objective").

Token used in the message:

{copilot=temp__creative
The customer stated that they want to rent the car for {var=number_of_days} days. Consider the following pricing rule:

Up to 3 days: R$100 per day.
From 4 to 7 days: R$90 per day.
More than 7 days: R$80 per day.

Return the total rental amount based on the number of days, explaining the reason for the applied amount.
}

Send Message block.

Testing the AI behavior.

As shown in the image below, the token was correctly interpreted and calculated the total rental amount based on the defined rates.

In addition, it is possible to include other tokens within the {copilot=...}, such as the lead's name ({contactfield=firstname}) or any other value, allowing the AI to access this information and further personalize the response.

Chatbot Support.

If you want to add a fallback to identify possible errors when using the token copilot, just insert a | before closing the braces and then add an error message. For example: {copilot=Your prompt here|Error identified, AI configuration problem.}