Our mission is to provide the information our customers really want. We face public transportation challenges, mainly in Japan. In Asia, it is becoming difficult to hire staff due to the declining birthrate and aging population. It will get worse and worse. Passengers can now get information from smartphones and boards at stations, but those who want other information cannot use it. We solve this big problem by providing the information that each person really wants.

What it does

We developed the communication tools which combines digital signages at stations and call center. This allows passengers to communicate with specialists even if we don't have staff on site. The content of the conversation is analyzed by AI. Frequently asked questions are displayed on digital signage to make it easier to obtain information.

How we built it

Our system is built up of two front-end web apps. One is the Admin App for call center staffs to use and the other is the kiosk app that runs on digital signage.

We used zoom-video-sdk to enable communication between these two applications. In addition, all backends are using OCI's PaaS service, which constitutes a microservice architecture. We also used OCI's AI service for text analysis and summarization of conversations.

To make sure that the Kiosk app runs on digital signage, we built our own digital signage device with a Raspberry Pi.

Challenges we ran into

Because of the structure of the two applications working together, it was difficult to test them when connected. Moreover, since the team members were from various nationalities and locations, it was also challenging for us to communicate with them.

Accomplishments that we're proud of

  • To have built almost all of OCI's PaaS components that can be used in a serverless environment.
  • All event-driven implementation from Zoom Cloud Recording to storing the transcriptions in the NoSQLDB.
  • Having also implemented the necessary parts of the actual product, such as API token authentication and on device testing.
  • Above all, we brought together engineers from railway companies, airlines, and various countries to create something that will benefit the world.

What we learned

  • How to create a backend with OCI
  • How to implement Zoom Video SDK and use event hooks
  • How to integrate Firebase Auth authentication into the OCI APIGateway
  • How to create microservices with serverless architecture
  • About Material UI

What's next for decide this later

Interested in Zoom's Contact Center SDK. Would like to change the call portion from HTTP to Websocket for more real-time processing. Need to organize our OCI policy settings to improve security. Not able to accomplish the automatic FAQ generation part this time, so as soon as the ChatGPT API is released, we would like to consider a way to do this.

Built With

Share this project: