RoadSpice

Inspiration

After reading the 123526th article on the perfect road trip from LA to New York, the idea for an AI powered app emerged.

What it does

RoadSpice takes as input a desired start and end location, as well as how much the user wants to veer of the shortest path to construct a scenic trip.

How we built it

The frontend was constructed using React, Tailwind, and Typescript. The user input gets fed to Google's Gemini 1.5 Flash. The LLM returns a summary of key points along the path which are then used as input for the Google Maps API to visualize the path.

Challenges we ran into

After multiple exposed API keys, hours debugged on git merges, handfuls of hairs pulled out to get maps API working, and fine tuning Gemini prompts, the biggest challenge was picking a good name for the project.

Accomplishments that we're proud of

As first time hackathon competitors, successfully finishing everything we set out to do is out biggest accomplishments.

What we learned

We learned new skills about using the Gemini API, Google Maps API. We refined our skills with react and tailwind for creating a UI. We also learned how to work as a team effectively, work under time pressure, and ingest caffeine while typing.

What's next for RoadSpice

We were experimenting with using FetchAI to add restaurant, hotel, rest area, and gas station recommendations along the way. We were unable to finish working on these agents, but we plan on implementing them after the hackathon to polish this project.