BKK Airbnb Ai Chatbot

Inspiration

Creating an intelligent chatbot has always been a challenging task. However, with the advent of advanced NLP models like ChatGPT, it has become not only feasible but also enjoyable to work on such projects. The availability of these cutting-edge models has revolutionized the way we develop applications, opening up new possibilities and opportunities. Being a part of this technological advancement is incredibly thrilling and inspiring.

What it does

The application will serve as a helpful assistant, aiding users in finding suitable Airbnb accommodations in Bangkok, one of the most appealing destinations for travelers worldwide. Its purpose is to simplify and enhance the process of discovering and selecting the ideal place to stay.

How we built it

Our application gathers Airbnb listing data that corresponds to user preferences, including location, price, property type, amenities, accommodation size, reviews, ratings, and more. We ensure the cleanliness and quality of the collected data. To enhance its efficiency, we employ Cohere's advanced technology to embed the data. Subsequently, we leverage Pinecone's indexing and storage capabilities to efficiently store and retrieve the embedded data. Additionally, we utilize Langchain as an agent to determine appropriate actions and responses based on user queries. Langchain is empowered by the powerful LLM model, such as gpt-3.5-turbo, enabling intelligent and context-aware interactions.

Challenges we ran into

Limited Support Credits: Since we entered the competition relatively late, we were unable to obtain support credits from platforms like OpenAI, Hugging Face, etc. Consequently, we had to work with minimal configuration resources, relying on free tiers and other cost-effective options.
Building our first AI Chatbot: Developing an AI chatbot from scratch posed a significant challenge as it was our first endeavor in this domain. Starting with limited knowledge and experience, we had to learn and adapt quickly to transform our concept into a minimum viable product (MVP). Despite the intense learning curve, the journey was rewarding and enjoyable.

Accomplishments that we're proud of

Despite this constraint, we strived to optimize our implementation to deliver a MVP application.

What we learned

A lot. Integrating different technologies and platforms, such as Cohere, Pinecone, and Langchain, posed its own set of challenges. Ensuring smooth communication and compatibility between these systems required thorough testing, debugging, and fine-tuning.

What's next for BKK Airbnb

Feature Expansion: Consider adding new features to enrich the user experience. For example, adding the booking process.
Localization and Global Expansion: Consider expanding the application beyond Bangkok to cover other popular destinations worldwide.
Enhanced Personalization: Refine the recommendation system by leveraging user feedback, ratings, and historical data to provide even more personalized and accurate suggestions.
Mobile Application Development: Extend the application's reach by developing a mobile version for iOS and Android platforms. This will enable users to access the application conveniently on their smartphones and tablet.