Equip your bots with trusted generative AI

In addition to its other features dedicated to bots and conversation automation, iAdvize lets you use trusted generative artificial intelligence to handle conversations on your own. This article explains how bots with this AI work and presents the associated use cases.

 

Principle

By using trusted generative AI, you enable your bot to handle conversations completely autonomously : it generates its own answers based on visitors' questions and the information available to it, rather than using the script's pre-written answers.

 

However, you retain control over what the bot says:

  • by deciding which sources of information (AI knowledge) it will draw on to answer your visitors' questions.
  • by programming personalised actions in the event of certain situations (technical error, no response, prohibited word or expression, request to speak to a human).

 

Generative AI is implemented using a dedicated map of a bot's scenario. You can therefore combine it with other maps to personalise your scenario according to your conversational strategy.

 

The different cases of use

Product information

To respond to this use case, you can feed a bot with the data you already have on your products. The bot will be able to understand your visitors' questions about the products and answer them, within a conversation, by drawing the answers from the data provided.

 

Example : your visitor wants to buy the SuperAspi2000 hoover. He has questions about this particular model (price, technical characteristics, repairability score, etc.). Based on the product information provided, your Gen AI bot is able to answer your visitor's questions independently and in natural language.

 

Product Discovery (private beta)

This product discovery feature enables your Gen AI bot to suggest products to your visitors that match their search criteria :

  • it can adapt to requests made in separate messages,
  • it can ask visitors for more details if they lack the criteria to find the right product,
  • If your visitor's criteria change as the conversation progresses, it adapts by offering products that meet these new needs,
  • it suggests 1 to 3 products to your visitor in the form of a carousel; each product is accompanied by a short paragraph to make comparison easier.
  • it then checks that its proposal matches your visitor's search. If this is not the case, it will suggest a new product until it satisfies your visitor's search criteria.
  • if no product matches your visitor's search criteria, it invites them to change their criteria to find a more suitable product.
  • finally, if it detects that it is unable to find the product requested, and/or that its recommendation poses a risk to the company, it automatically transfers the conversation to a human respondent.

 

Example : your visitor is still looking for a hoover. They don't have a particular model in mind, but they'd like a cordless model with a specific budget. Based on these criteria, the Gen AI bot draws on the product information available to it to offer your visitor a range of choices.

 

FAQ

Your visitors no doubt often want to get a quick answer without having to consult your entire FAQ. So you can feed your bot this rich source of information. Thanks to this feature, your bot will be able to respond autonomously to all the FAQ-type questions for which you have provided it with information.

 

Ex: Your visitor has now found the hoover that's right for him. But he still has questions about delivery and after-sales service. Your Gen AI bot will be able to answer them autonomously on the basis of the FAQ information you provide.

 

Configure your bot

Find out in this article how to set up your bot, and then how to provide it with your information using "AI Knowledge".