Inspiration

We are travelers and explorers at heart. We believe that our time should be spent discovering new places and experiencing adventures, not bogged down by planning. Pocket Travel was born from our desire to make travel effortless and inspiring, so you can focus on the journey, not the logistics.

What it does

Simply tell Pocket Travel where you want to go, how many days you’ll be staying, and when. Our engine will instantly plan the perfect itinerary, considering factors like time, season, and personal preferences. It handles the tedious work of researching reviews, comparing ratings, and checking opening hours—saving you countless hours of planning.

How we built it

We identified key pain points through user journey analysis and split the development into a front-end interface, back-end orchestration, and an AI-driven decision-making engine. Using advanced prompt engineering and scraping SERP API data from Google, we crafted an intuitive solution that makes trip planning easy.

Challenges we ran into

One of the biggest challenges was choosing the right AI model that balances speed, creativity, and accuracy. We needed something that could quickly generate complex itineraries while keeping personalization and precision intact.

Accomplishments that we're proud of

We’re proud of creating a system that generates perfectly timed itineraries within seconds. Instead of a basic text-based chatbot, Pocket Travel offers contextual, multimodal information, making it extremely user-friendly and highly efficient with minimal input from users.

What we learned

We learned the strengths of Mistral AI’s models, especially Pixtral for multimodal prompts and the Large 2 for logical planning. Our experience taught us that coding copilots significantly boost productivity, and effective prompt engineering leads to better interactivity, allowing the engine to handle edge cases and continuously engage users until all trip details are finalized.

What's next for Pocket Travel

Next, we plan to allow users to modify specific parts of their itinerary and have the model reselect just those elements to further personalize the trip. We’ll also enhance user preferences based on past interactions and integrate GPS and transport data to better factor in proximity and travel time for more realistic plans. To improve decision-making, we will incorporate data from multiple travel agencies and sources to provide a more holistic view.

Built With

Share this project:

Updates