Avian

An AI powered travel assistant that generates itineraries tailored to user preferences cutting travel planning from hours to mere seconds.

Comment

Inspiration: The Traveler’s Dilemma

Trip planning can be time-consuming and overwhelming, requiring research on local activities, dining, etc.

Families especially those with dietary restrictions and accessibility needs require a variety of accommodations in their day-to-day activities.

What it does

Avian tailors every itinerary to user preferences, including desired experiences, trip length, dietary restrictions, and accessibility accommodations. Delivers unique, context-aware plans for solo travelers, couples, and families. Eliminates hours of research by generating comprehensive, day-by-day plans in seconds. Helps travelers focus on the experience, not the logistics. Users can refine inputs to adjust the itinerary dynamically.

How we built it

Created a simple UI using javascript and styled with CSS, integrated OpenAI's API and used prompt engineering to refine the generated output. We first used persona technique in order to guide the AI's responses as a travel assistant. Then we narrowed down and formatted the outputs into a detailed time-based itinerary by using the few-shot technique. Lastly, we used iterative refinement by adding several user inputs (trip length, desired experiences, dietary restrictions, accessibility accommodations) to fully customize the experience for the users.

Challenges we ran into

  • We were unsure of what features to include other than just the itinerary generation.
  • The buttons were not aligning correctly in the center like we wanted.
  • The formatting in the email format was not properly formatted and had
    tags.
  • API set up had several errors along the way and required a lot of troubleshooting.

Accomplishments that we're proud of

  • Learned how to use OpenAI API effectively.
  • Dived into prompt engineering and learned several techniques (Persona, few-shot, iterative).
  • Working as a team effectively.

What we learned

  • How to integrate the OpenAI key to effectively generate relevant data given the user's input.
  • Prompt Engineering: Few-shot technique, persona technique, iterative refinement

What's next for Avian

  • Integration with major booking platforms for facilitated planning.
  • Accurate travel time adjustments based on arrival/departure times.
  • AI-powered packing lists and budget management tools.
  • 'Edit' option in order to swap out specific suggestions.

Built With

Share this project:

Updates