Inspiration
Our team was comprised of graduate students researching AI, and thus we were naturally drawn to this project! It's concrete terms inspired us; we were excited to explore new technologies and confident we could deliver value to AutoNation's business in short time.
What it does
Our app is a chatbot. The chatbot helps enhance customer experience at AutoNation by responding intelligently to customer queries, increasing customer satisfaction and reducing employee workload, such that staff can focus their time smartly on more complex customer interactions, and reduce time spent on mundane tasks-- a win-win.
How we built it
This app was centered on Google's Dialogflow natural language platform, Google Cloud, and Twilio for SMS messaging. We used API's constructed from these platforms as well as webhooks made in Python's Flask framework, hosted on Heroku and Google Cloud to connect together our application. For the store information details provided to us, we directly sent it through the backend server, whereas for more general customer information, we trained the Dialogflow model to generalize well.
Challenges we ran into
We initially sought to leverage Python NLP libraries for this project before pivoting to use Google's Dialogflow, which proved extremely convenient, perhaps most especially for its ability to integrate easily with other parts of our application. We realized that by using Google's Dialogflow, we could easily make this chatbot available across many platforms such as Facebook Messenger, Telegram, Slack etc.
Accomplishments that we're proud of
We are very proud to have completed a working, robust app which satisfies AutoNation's requirements in such a short time! Our app is scalable and flexible to work with various user interfaces, so far we support sms, Google's dialogflow chat, and Facebook Messenger, which may potentially enable the inclusion of payment and additional functionalities in the future!
What we learned
We learned simple, flexible solutions are highly desirable under time constraints; it is important to know at which scales and time allowances marginal performance gains are preferred.
What's next for AutoAI
We would like to add support for more languages, and generally improve on our model's language processing to increase the chatbot's functionality. As our platform also enables non-technical staff and stakeholders to contribute to its development, we imagine it will continuously improve and add value to AutoNation throughout it’s life! We're excited to improve based on feedback from AutoAI's users!
Log in or sign up for Devpost to join the conversation.