Hardly anyone likes to communicate with a virtual assistant in a call center or chat.
However, automating responses to a growing influx of inquiries is too profitable to ignore, so companies are looking for a balance: the quality of the customer experience on the one hand, and the cost of customer service personnel on the other.
We have developed our own approach to implementing AI and chatbots, which achieves 30-60% automation without a drop in customer satisfaction in terms of CSAT.
Realistic Current Capabilities of Conversational AI Technology (NLU)
It is important to understand what commercially available AI can and can’t do.
Can
- Match a client's message to one or more predefined topics (in some way it is an “understanding”)
- Extract data from the message (for example, time, date, city)
- Learn, but only with the help of a person
Can’t
- Compose their own answers
- Build a consistent dialogue
- Learn completely on their own
This is why attempts to make a robot that imitates human communication most often fail
Our Solution: AI + Chat Apps
So, the AI can't construct a dialogue on its own, but it's good at identifying the basic topic of the question from the client's free-form message. This is where what we call “chat apps” come to the rescue. You can compare them with mobile apps, but they’re located inside a chat. They are controlled by buttons and other interface elements.
For example, Apple Messages for business has the ability to send an Apple Pay button to the chat. Whatsapp Business has a build-in “add to cart” feature. Just imagine what opportunities this opens up in terms of promo mechanics when interacting with clients.
Summary
The idea is simple: the AI identifies the mechanic and launches the chat app.
The customer doesn’t need to continue the dialogue with words, which the bot might not understand. They can simply press the necessary buttons or send the necessary information to the bot and solve their problem.
If the AI finds multiple topics with a >90% recognition rate, it will offer all the scenarios it finds and the client can select the appropriate one with the press of a button.
For these tasks, we are making Brandy.im - the brand interactions platform. It combines chat apps development studio and conversational AI. These bots can query and exchange data with other systems. Thanks to this, it is possible to customize them for the main topics of the particular business. Resulting in a smooth customer journey.