When mapping out your prompt there are 3 main components to think about:

1. The main guidelines, that help you design an overarching base from which the AI will follow. We call this the prompt structure.
2. Next is adding in some emotion or how you would want your AI chatbot to respond - somewhat adding a humanised element of conversation. This is called the sentiment design.
3. The final component is where you can add your own brand tone, so the AI will respond how you would normally talk to your customers. This is called the tone delivery.

Here are 2 basic examples of the prompt structure. You can see that we're setting the foundation to how we want the AI to respond in a way that is helpful and encourages the customer to ask more questions, that can help resolve an issue faster. Your prompt can be longer or shorter. For example we could have a simple 3 word prompt, that makes the entire AI multilingual and will respond to customers in any language. 

Adding the right tone of voice is important if you want the AI assistant to respond in a tone that your customers are already used to. Or you can quickly change it out for a more seasonal tone for example: 

1. It's Halloween 🎃, and you want to switch out the conversation to a more spooky response that will freak your customers out.
2. Well we know Christmas brings in the ecom revenue like there's 'Snow' tomorrow, so let father xmas 🎅 take the reigns.

Setting your AI prompts will take just a few seconds. You can always test and fine tune any of the variations until your AI responses are how you like them.

1. Open the GitBook action and navigate to where you added the AI response field.
2. Next is to type in your prompt structure, followed by the sentiment then add your tone. At the end of the prompt then add your AI response field.

Note: You do not have to add them all, for example if you just wanted to add the tone delivery, you can do so.