Team Members: Atishay Jain (User ID: _YIm3GVnOhrzgJyq5i-R5A) Naisheel Shah (User ID:6PFWftoa05y8OyyVXFguVg) Ojas Naik (User ID: b76SqKgJaGcidZPgsR-vNg) Samarth Ghulyani (User ID: BkbmRKhLTsnCKOvhHmCXhw)
We think that everyone, regardless of age, has come across the age-old dilemma- "I want to go out for the long weekend, will my sibling/partner/parent/friend join me? I don't want to go alone. Even if I do, how do I plan it? I have no idea about the place. I'm clueless, I guess I will just watch Family Guy and call it a weekend." This is where Trippy comes in. We solve this problem by giving the users all kinds of choices to go on their dream vacation; from activities, hot tourist attractions, trendy cafes, monuments to hotels, flight bookings and so on. And we do this while keeping in mind the user's likes and dislikes. Intelligently preferring one over the other with minimal load on the pocket. Trippy not only plans it but finds other users on the platform who may share the same interests as you do and connects you with them so you can plan the best trip ever.
Essentially, Trippy is a hub of travel agents. Think of it like this- you want to book a flight? You go to skyscanner.com. Do you want to book hotels? You visit Booking.com, only to find a million hotels that may or may not meet your needs. Going to a new place/country? You start web scraping through Google and Reddit to find those spots that tickle your tourist persona. For a 2 day vacation, you end up tediously surfing the web for hours and end up with more questions about the destination than when you started planning. With Trippy, however, you book a flight with the help of our flight agent, and you book hotels with multi-cuisine restaurants as you have always wanted to try using our 'fetch hotels' agent. We have 7 unique functionalities that are crucial to trip planning. It doesn't end here, we take all this information and find the perfect getaway for you. Giving you a detailed vision of what your vacation looks like. So you don't have to worry about things happening unexpectedly or because you weren't prepared. We take care of it all.
At the heart of our project are several "agents" that intelligently answer questions that are relevant to the function they can perform. We made these agents using FetchAI's masterful framework. Trippy is a product of a ton of brainstorming. We started off by thinking about exactly what people look out for when they're planning a trip. We decided to develop 7 agents from scratch to curate a satisfactory itinerary for the user and these agents gave us enough data to see what users like and 'connect' to other users on the platform who share similar tastes. We do this by creating a dynamic embedding into Trippy that checks the user's itinerary with others in the database and calculates a similarity index. The seven agents are- FlightAgent, HotelsAgent, ActivityAgent, RestaurantAgent, TopDestinationAgent, IteneraryAgent. The functionality of each agent is as follows: -- FlightAgent: Uses user's convenient date, time and destination they want to go and suggests flights from their location to destination. This agent also can manage preferences, for example, if the user wants flights with certain stars on Google reviews or if the airlines offer in-flight meals etc. -- HotelsAgent: As the name suggests this agent is responsible for fetching hotels as per the budget (if mentioned), or any preferences there may be. -- ActivityAgent: Sees the number of days the user wants to travel and sees the interests like hikes, concerts, festivals, shopping, sightseeing etc. and suggests what can be done in that # of days. The parameters in all of these agents depend totally upon what the user mentions while chatting with Trippy. -- RestaurantAgent: Suggest famous, cost-friendly restaurants as per cuisine preferences (vegan, fine dining, fast food etc). -- TopDestinationAgent: Returns a list of must-visit places in the destination spot -- IteneraryAgent: Now that all the information about users' vacations is with our agents, we compile all of this information and create an itinerary for the user. A few quality-of-life features were added like giving users the ability to see similar itineraries, their ratings and reviews made by users of the platform who followed this itinerary for their trip.
Getting through the FetchAIs framework is the toughest part of this project by far. This technology was completely new to us because we were under a strict time constraint. We took the first day to figure out how a single agent works. We experimented with their agentverse and decided to make a dedicated UI for this project as the application of these agents to this project seemed ideal. We essentially made a specific application for the idea of agentverse, in which we have 7 agents that can interact/chain/talk with each other to compile a meaningful output. Another lesser major challenge we faced was integrating our React UI with the agents. We learned about fast APIs and setting up continuous calls between agent-proxy-ui. Where proxy was used basically as a bridge between agent and UI to talk to.
We developed an intelligent fully functioning network that assists people of all kinds to make the best out of their vacation. A major motivation for this project also was the fact that a lot of people in today's world do not get the time for themselves and save for years to go on a vacation. Trippy makes sure that all preferences/choices of the users are maintained without compromising the experience. Another is that our matching functionality has huge potential. It lets people who are shy or introverted, who like to travel form meaningful interactions with people who share these interests. We think that Trippy works as a self-fueling vehicle in the outdoor community that motivates people to be more active and social.
This was the first hackathon experience for our team and we learned a ton of things with this project. We learned to manage tech stacks on the go, learning through documentation, and lots of prompt engineering to solve bugs that were never seen before. Conversing with other teams inspired us to do better and make a better product that could potentially reach people in real life.
-- Enabling booking/ticketing/ordering for flights/trains/buses -- As mentioned above the social aspect of Trippy is humongous. Our next step for trippy would be to add a grouping functionality, which would work kind of similar to matching 2 users, except here we plan an itinerary for a group outing where each member has a say and decides on a solid vacation plan using our robust Travel agents.