frontend
dashboard built with figma
video chat backend on metered.ca

Inspiration

Customer service is a top-tier priority for many companies. Gaining analytics into communication between a customer service representative and the client can help companies identify areas of improvement as well as can speed up the process of addressing the clients' issues.

What it does

Basically, we developed a web app that is a portal for organizations to extract data such as sentiments, topics of the conversation, and bulleted summaries of the conversation that took place between the customer service representative and the customer. The data is then visualized for further interpretation. link

How we built it

We used React to build the frontend component, google Firestore NO-SQL database, and flask for the backend. For the text analysis, we implemented Assembly AI.

Challenges we ran into

We tried various approaches such as Mux API for building a video conferencing App. However, we could not get it to connect with the Assembly API. Also, we ran into issues of finding customer service phone calls audio files for testing the accuracy of the Assembly AI's model different components.

Accomplishments that we're proud of

We made our own web socket for communication. YAY!! :')

What we learned

We learned a lot about real-time transcription using web sockets. We learned how to use Firestore in production mode.

What's next for Customer AI

In the future, we will like to implement Customer AI in different languages through Assembly AI. For each speaker, we will like to have their own preferred language "subtitle" in the meeting space.
Also, we will like to add a bot that can communicate with customers directly. Another one, we can add slack or even notion integrations.