Google’s mission is to organize the world’s information and make it universally accessible and useful. To make powerful computing tools accessible to all, Google offers the Google Cloud Platform (GCP) suite of cloud computing products. This platform includes products that assist in data management, artificial intelligence, and application security.

When a GCP customer encounters an issue and contacts support, manually collecting error logs and project details is time-consuming. Additionally, many problems can be quickly fixed by pointing the customer to public documentation about the product. These types of problems take significant time and slow down customer support for other users.

Our Self-Service Support Chatbot for Google Cloud is a suite of tools, including a customer support chatbot that automatically gathers information about errors in order to identify and provide relevant documentation available on the web.

Our chatbot parses and analyzes a user’s question using natural language processing (NLP) and provides relevant information that might help with their problem. However, if the chatbot cannot solve the problem, the chatbot organizes the information it has collected and refers them to human support.

The user contacting human support then submits the information the chatbot has collected. This significantly cuts down on the support engineer’s time spent on error information collection and the overall time to issue resolution.

Our chatbot uses machine learning to improve document recommendations over time and tailors its resolution recommendations based on feedback from users.

Our chatbot is built using GCP’s Dialogflow and Google’s NLP API. React serves the front end for the chatbot and Flask hosts the back end on Google App Engine.