This project is still ongoing! >ᴗ<
I will be updating this case study periodically.
Try it out!
cmuGPT was recently deployed and available to check out! We are currently working on the front-end!
01 PROBLEM & CONTEXT
CMU has A LOT of websites & webpages…
I love Carnegie Mellon University with all my heart, but like many large organizations, CMU has many websites with complex pages, making it hard to quickly find relevant information.
This project was created as a team within Scotty Labs, a community of passionate creators within CMU. This project was created and is led by Thomas, a CS student at CMU.
02 RESEARCH & IDEATION
A chatbot for cmuMaps
cmuGPT was not always cmuGPT. Initially, Thomas wanted to create an NLP integration into cmuMaps, which would allow streamlined campus navigation for students. The use cases of the chatbot were threefold:
01 Location Search Descriptions
Students can locate certain spots / classrooms on campus by description
02 Alternate Route Setting
The NLP integration looks for fun events happening on campus, and creates alternate routes to the destination
03 Search for Specific Foods
Using our dining API, we can scan menus and search for specific foods at specific times and locations
Thinking bigger…
While an integrated chatbot was cool, I had a strong urge to make this project something bigger and more universal. During one of my talks with Thomas, I suggested that it'd be pretty fun if we web scrape all CMU-related websites to create a standalone chatbot. And thus, cmuGPT was born.
03 BRANDING
Something playful & warm…
04 REFLECTION
What I'm learning…
Advocating for Design
When I first joined cmuGPT, there was heavy emphasis on the back-end processes. I found that there often wasn't a clear vision of the product or the user flow. In some ways, I think that's why me an Thomas make such a good team. We have different strengths and perspectives we contribute to cmuGPT
Communicating with Developers
I'd often have to pester Thomas and our developers about the end product, coming up with example use cases and checking to see if they were feasible.
