Inspiration
A common struggle with shopping for a car is not knowing where to begin and being overwhelmed with choices and information. Our goal was to match a customer with their dream car by having them talk to a virtual agent that would showcase the top 3 best options for the user during their conversation. This meant the user can be as specific as they want and ask the agent a 100 questions or keep is brief and only ask 1.
What it does
Take in input from the user and guide them with questions on what they look for in a car. The bot takes into account a user's interests and personalities to come up with the perfect match!
How we built it
Front end
- NextJS
- React
Back end
- NodeJS
- PostgreSQL
- Prisma
- Python
Deployment
- Vercel for our web application
- Heroku for our worker python web scraper
Challenges we ran into
For the first two hours of the hackathon we delved into an idea that quickly ran into a dead end. After we found Toyota's challenge we needed more information, so we approached Toyota's booth to further inquire about what they would like to see. Another challenge we faced was not taking sufficient brain breaks which hindered our productivity.
Accomplishments that we're proud of
- Fully integrated chat bot ui with special system prompt instructions.
- Deploying our website under a custom domain
What we learned
- Prisma syntax and commands!
- Working with OpenAI's api
- System prompting
- Creating dynamic web pages
What's next for yepai
- Continue polishing the site to its full potential post hackathon
- Run user studies and gather feedback to improve our UI and ai prompting
- Develop a voice activated mode to mimic more of a phone call like scenerio
- Implement showcasing used car listings
Built With
- heroku
- nextjs
- postgresql
- prisma
- python
Log in or sign up for Devpost to join the conversation.