Accessing and integrating property data securely was the first challenge. The chatbot needed structured data for matching while maintaining connection through REST APIs.
The bot was required to respond only with property data from the website. We restricted and filtered responses to ensure no external data was used.
We optimized GPT token usage to reduce cost and maintain efficiency. The model was trained to extract intent with minimal processing.
We used APIs to fetch reviews from Google and TripAdvisor, performed sentiment analysis, and tagged insights like “quiet rooms” or “clean bathrooms.”
Combining user input, property data, and review sentiment into a single matching flow required careful logic design for accurate recommendations.
We built a simple, lightweight web interface to test chatbot interactions, show results, and validate AI behavior before full integration.
The GPT-based engine interprets user travel queries and extracts structured details like location, month, and preferences automatically.
We used real reviews and metadata to match user needs accurately and justify recommendations based on guest feedback.
By refining prompts and training patterns, we reduced token usage while improving accuracy and maintaining performance.
Python backend was connected securely with WordPress via REST API for smooth communication and controlled data flow.
Scraping scripts analyzed reviews to extract sentiments and keywords, enriching property data for smarter recommendations.
We created a clean test interface where users could chat, test scenarios, and see live property recommendations in real-time.
Used for AI, backend, and data processing.
Connected JS frontend to WordPress securely.
Handled AI intent and entity extraction.
Integrated chatbot inside the site
