marco.ai

Meet marco.ai: Your personalized travel companion, using AI for personalized recommendations and real-time insights, making travel seamless and unforgettable. swipe, and go!

Comment

Inspiration

Traveling can often be stressful, with countless variables to consider such as budget, destinations, activities, and personal interests. We aimed to create an app that simplifies travel planning, making it more enjoyable and personalized. Marco.ai was inspired by the need for a comprehensive travel companion that not only provides an engaging way to 'match' with various trip components but also offers real-time, personalized recommendations based on your location, ensuring users can make the most out of their trips.

What it does

Marco.AI uses your geolocation on a trip to find food and activities that might be of interest to you. In comparison to competitors like Expedia and Google Search, Marco.AI isinputted with personalized data and your present location and provides live data based on your past preferences and adventures! After each experience, we ask for a 1-10 rating and use an initial survey to store a profile with your preferences.

How we built it

To build Marco.ai, we integrated You.com, Groq Llama3 8b, and GPT-4o APIs. For the backend, we utilized python to handle user data, travel plans, and interactions with third-party APIs in a JSON format. For each experience, our model generates keywords relating to a 1-10 rating of the experience. Ex: a 10/10 for beach would have keywords like "ocean, calm, relaxing." For the mobile app frontend, we chose React Native to connect to our model output and present an easy to use interface. Python's standard libraries for handling JSON data were employed to parse and save AI-generated recommendations. Additionally, we implemented functionalities to dynamically update user ratings and preferences, ensuring the app remains relevant and personalized.

Challenges we ran into

As first-time hackers, we definitely ran into a few obstacles. Combining multiple APIs and learning how they work took a while to figure out and implement into our model. We started off by using mindsDB and trying to utilize a RAG model with You.com. However we realized that for our purpose, we didn't need to use a largescale model management and platform and decided to move towards using prompt engineering. Engineering our prompts for GPT-40 was a back and forth process of learning how to properly utilize the AI to give us our output in a formatted way, making it easier to parse.

The most challenging aspect for our team was frontend design. This was our first experience with app development, and the learning curve was steep. We are happy that we are able to provide a functional prototype that can already be used by people while they plan a trip!

What's next for marco.ai

In the future we hope to bring this model to life with payment integration and a feature to be able to swipe through and save different elements of your trip: hotel, flights, food based on your interests and budgets then pay at the end. Additionally, we aspire to transform Marco.ai into a social platform where users can share their past vacation experiences, likes, dislikes, and recommendations, creating a vibrant community of travel enthusiasts! Marco aims to pioneer a social app focused on encompassing travel experiences, filling a gap that has yet to be explored in the social media landscape.

Share this project:

Updates