Kuprevicha Str. 1/1, Suite 923a

Minsk, Belarus

Business trip to Sweden
Feb 16, 2018

Business trip to Sweden

We`ve just returned from a business trip to Sweden, where our key partners, Hostit, live and work. We make some important decisions with them. They are ideological inspirers, we are quality performers. One of Ticketbird’s projects has already developed and begins the life of the finished product. This is an online platform based on SaaS (Software as a Service) which offers customers solutions to their questions related to the goods and services acquisition (or manifestation of interest) of companies from any sector of the economy.

Do you remember pop-up chat windows on different sites? They are more annoying than helping, because they are answered for a long time or go unanswered at all. Now we are making a system that will make such chats much more useful.

Ticketbird is a hybrid of the client chat on the site and smart search. This “artificial operator” takes questions from customers and knows the answers to most of them as . Responses faster than the live operator.
Here is algorithm of Ticketbird:

  • The client writes to the chat and the smart bot is included in the case
  • The smart bot reads the client’s question and answers to the best of his ability.
  • The smart bot finds information in the database of their own knowledge
  • And prompts the desired product in the catalog or sends it to a live operator.

The main difficulty of developing such a system is to teach a machine to understand a person. After all, people are all different and they ask questions in different ways: in different forms, in different words, with misprint. We have written a self-learning algorithm that recruits vocabulary as we communicate with people. Just like a child.

The freelancers from Ukraine and Russia who had been working on the project for three years tried to solve this problem with us. After the project had been handed over to us, we had to redo everything in the code and bring it to mind. We made a smart catalog search in two weeks.

Now the project is almost ready for launch. We have already new customer – large catalogs of auto parts. We will follow our clever algorithm, teach it and retrain. As soon as we get enough knowledge about the issues it will be able to release the load of operators by 3/4. Customers will receive a response faster and companies will be able to reduce the cost of client`s support.