Remember chat popups on various websites? They tend to annoy more than they help, because you may wait for your response too long, if you’re lucky enough to get one at all. We are now working on a system that will make such chats a lot more useful.
Ticketbird is a hybrid of website-based client chat and smart search. It is a sort of “artificial operator” that receives questions from customers and knows answers to most of them. It usually answers faster than a human operator.
It works like this: a client sees the standard chat window on the website of a company, which has Ticketbird deployed. If the client writes a message, the smart bot is activated. It reads the client’s question and offers the best answer it has. It finds information in its own knowledge base, suggests a proper product in the catalogue or sends it to a human operator.
The main difficulty in developing such a system is that you need to teach the machine to understand human requests. After all, people are all different and they ask questions differently – in different ways, using different words, with misprints, after all. We have developed a self-training algorithm that expands its vocabulary as its communicates with people. Almost like a child.
The project is almost ready for launch; we already have our first customers – large catalogues of automotive parts and a sensor manufacturer. We will keep an eye on our smart algorithm, teach it and retrain it. As soon as our bot has accumulated the necessary amount of knowledge, it will be able to reduce the burden on human operators by 75%. Clients will enjoy a faster response and companies will be able to cut their customer support costs.