Inspiration
We see people all about "the grind". But then they forget about hanging out with people, eating properly, do fun things...or forget to touch grass. Most of these people shut themselves out so talking to them isn't a solution. So if we can't bring them to the solution, we'll bring the solution to them, on their smartphones. TouchGrass is going to be their personal AI assistant and help them plan their lives out better. Make them plan their work-life balance properly.
What it does
TouchGrass takes in the user's steps per day, weekly schedule and their daily geolocation to determine if they have been working too hard and need to "touch grass". It gives recommendations such as places the user can go, or if they need to walk more and even give mental health advice based off of a database of mental health books.
How we built it
Fireworks.ai -> Mistril 8x7b LlamaIndex -> RAG to create a vector-store index of mental health-related books to give users relevant advice with context.
Challenges we ran into
Google Authentication to get geolocation and step counts was troublesome and we used a lot of time on that.
Accomplishments that we're proud of
- Able to create a chatbot that actually reads the latitude and longitude of the user location and recommend points of interest nearby
- RAG to have an LLM intelligently call google geolocation API to find the users location and call google calendar API to recommend event ideas through function calling.
What we learned
- How to use Fireworks.ai and LlamaIndex, vector-stored indexes
- How to use Google Fit API, Google Calendar and React Native
What's next for TouchGrass
- Better UI for better ease of usage amongst users
- Speech to text functions to add realism
Built With
- fireworks.ai
- javascript
- llamaindex
- react-native
Log in or sign up for Devpost to join the conversation.