South western railway and Angel trains
What it does
We provide a single easy-to-use application to allow customers to voice their questions during their commute. We respond automatically via an AI assistant which leverages a knowledge base to answer the most common questions, improving customer satisfaction while reducing the number of issues directed at staff members.
Live issues raised are formalised to be shown to the staff. This allows delegation to the best suited staff member even if it is the closest. This allows for issues to be resolved faster while preventing other staff members from being overloaded by messages.
The final user is station and network control staff members which are able to see all the issues in their area however are kept from being overloaded by messages using a priority system.
How we built it
We used natural language processing to create a seamless experience for both the customer and staff members in the form of an AI chatbot.
Challenges we ran into
Docker containers running on ARM architecture.
Accomplishments that we're proud of
Creating a natural language model to understand questions from both the customer and staff members to improve communication. The application was built to work offline without the need for any third party cloud services
What we learned
First time using a mesh network facilitated by websockets and using NLP to create a seamless experience for both the customer and staff.
What's next for Ask Phoebe
Network wide communication monitoring. Modules for chatbots and analytics of queries. Steal Phoebes soul.