Our client is an India-based company that provides IT support and monitoring platforms for small to medium businesses in the US. The company has a US helpdesk department that allows them to serve US users with native speakers.
Business Challenge
Our client needed to address two primary challenges:
- As soon as a request came from a user, it takes a while to triage and assign it to the necessary engineer and that sometimes would cause problems with meting service-level agreements.
- Some requests did not require interaction with an engineer, and our client sought to process them without support agent involvement.
Project Description
Our client wanted to have a chatbot where users could input issues problem and chatbot would facilitate a solution by providing an article or running an automation action, and then assigning to an IT support engineer if the first steps did not product the desired solution.
SoftServe provided our client with:
- A chatbot service that is tightly integrated with Google DialogFlow and fine-tuned for the client’s specific needs
- An existing issue-tracking service integration that allows the triggering of resolution steps, that otherwise would have been made manually by a support operator
- Telegram-based UI component that is able to seamlessly join an on-going conversation and continue the chat as a support operator

Dialogflow was used for communication, and deep learning techniques were used for incident categorization. SoftServe overcame the challenge of understanding users’ problems from the free style text.
Value Delivered
The delivered chatbot solution allows our client to correctly categorize problems in 80% of cases and potentially resolve some cases automatically. This is estimated to result in a 20% faster incident resolution, as well as an estimated 10% of cases being resolved initially without support engineer involvement. The share of automatically resolved cases would grow based on analysis of processed cases and the automation amendments.