Croc robotized the first support line of employees "Megafon Retail"

2020: Robotization of the first support line

On August 12, 2020 IT- the company Croc announced robotizations the first support line of employees at offices and salons of network MegaFon MegaFon Retail(JSC). As the solution neural network the core of a dialogue system was selected AutoFAQ. artificial intelligence Will classify standard requests now and to automatically send them for execution. It will help to process 3 times more requests without staff increase, having reduced time of appointment of the contractor and terms of accomplishment.

Now more than a half of addresses on the first support line Megafon Retail of JSC can be processed without participation of the person. The intellectual core of a system understands freely formulated questions and unstructured addresses, working on sense, but not on a key word. It allows to determine different parameters of requests (a subject, a priority, urgency) with an accuracy of 98% and to appoint the contractor on them. Time for request processing of users was already on average reduced twice.

Qualifiers of addresses are constructed on the basis of neural network. For their training history of requests of users for the last 2 years on more than 800 categories was loaded into a system. At the same time the knowledge base of AutoFAQ samoobuchatsya due to machine learning technologies and artificial intelligence. Important advantage of the solution is an opportunity to check and adjust the trained models of qualifiers through a user-friendly visual interface without involvement of IT specialists.

Decrease in routine load of the staff of the first support line on average by 20% became one of results of robotization. It releases time of operators for other tasks. For example, they will be able to be engaged in data analysis, analysis of non-standard cases, including for the purpose of further training and development of qualifiers.

Appointment of the contractor, but also accomplishment of the standard request without involvement of the specialist can become the next step in development of technology not just.

The system of intellectual processing of addresses of users allowed us to increase considerably efficiency of the first line of technical support. Already at an implementation stage we reduced time of the first reaction according to the address almost twice. And in comparison with the same period of 2019 transmission time of an incident to the contractor decreased on average for 70 minutes. In plans - automation of all simple routes, and the first line at the same time will be able to help users online. The next stage – implementation of the chat-bot which will not just poll the user and classifies the address for transfer to the contractor, but also will propose the solution, and in case of a non-standard request will connect the employee of the first support line to the solution of a question, - Anton Zubarev, the head on management of service support of JSC Megafon Retail told.

Huge plus of the implemented system that potentially the quantity of models in it is unlimited and the database of standard requests will be able constantly to extend. It is very easy to manage training of models in AutoFAQ. Actually, we implemented the convenient tool which allows the employee to use achievements of artificial intelligence without special skills. It gives flexibility to business, allows to adapt easily to changes and to quickly scale the solution. Even if load of the first support line will become more, a system will quickly adapt to it. It allows to reduce costs of the company for a support service significantly, - Maria Ukolova, the director of the department of information technologies of CROC IT company comments.