AI-Powered Hotel Booking Assistant

Smart AI-powered hotel booking assistant built with Amazon Nova that allows users to search hotels, filter by price and rating, and get instant results through chat.

Comment

🏨 AI-Powered Hotel Booking Assistant using Amazon Nova

Project Overview

The Hotel Booking Assistant is a real-time conversational application that helps users search, explore, and book hotels efficiently. It leverages Amazon Nova foundation models for natural language reasoning and voice interaction, providing a seamless, AI-powered hotel booking experience.

Users can query hotel availability, prices, ratings, rooms, and bookings using typed input or voice commands, while all data is stored and managed in PostgreSQL for reliability and speed.

Why This Project Matters

Travelers often face fragmented hotel information and time-consuming booking processes. This assistant:

  • Simplifies hotel search and booking in real-time.
  • Uses AI to understand complex queries like "Show 4-star hotels in Karachi under 150 USD with available rooms next weekend."
  • Reduces manual effort by combining voice AI, reasoning, and database workflows.

How It Works

  1. Amazon Nova 2 Lite:

    • Processes user queries and reasons over hotel data to generate intelligent responses.
    • Handles complex multi-part requests, e.g., combining city, price, rating, and date filters.

  2. Amazon Nova 2 Sonic (Voice AI):

    • Converts speech input into text for natural conversation.
    • Enables hands-free hotel search.

  3. Hotel Tools:

    • Search by city, name, price range, rating, availability, and booking details.
    • Check room availability and retrieve hotel details dynamically.

  4. PostgreSQL Backend:

    • Stores hotel, room, and booking data.
    • Ensures fast, reliable query responses.

  5. Streamlit UI:

    • Provides an interactive interface with text input, voice input button, and real-time responses.
    • Displays conversation history for continuity.

Key Features

  • Real-Time AI Reasoning: Nova 2 Lite understands and answers complex queries.
  • Voice Interaction: Nova 2 Sonic enables natural speech input.
  • Multi-Tool Integration: Combines 9 hotel-related tools for rich data retrieval.
  • Database Integration: PostgreSQL ensures persistent storage and fast query results.
  • Conversation Memory: Maintains chat history for contextual responses.

Challenges We Overcame

  • Integrating Nova LLM with multiple hotel tools for seamless reasoning.
  • Handling voice input robustly with speech-to-text conversion.
  • Managing real-time database queries without slowing down the application.
  • Ensuring all modules (voice, tools, database, AI) work together smoothly.

Tech Stack

  • AI: Amazon Nova 2 Lite (reasoning), Nova 2 Sonic (voice)
  • Backend: Python, LangChain, Boto3, PostgreSQL
  • Frontend: Streamlit
  • Tools: Custom hotel search, booking, availability tools
  • Deployment: Cloud-ready Python application

Demo

What We Learned

  • Using foundation models like Nova makes reasoning over complex queries seamless.
  • Voice AI can significantly improve user experience.
  • Real-world workflows (hotel search + booking) can be fully automated using AI + databases.

Future Enhancements

  • Integrate Nova Act to automate booking confirmations and calls.
  • Add multi-city itineraries and recommendation ranking.
  • Incorporate multimodal embeddings for image-based hotel search.

Team / Individual: Muhammad Moaz
Hackathon Category: Agentic AI / Voice AI / Freestyle
Technologies: Amazon Nova, LangChain, PostgreSQL, Streamlit

Built With

  • amazon-nova-foundation-models-other-tools-/-technologies:-dotenv-(environment-variables)
  • bedrock)
  • boto3-(aws-sdk)
  • git/github
  • langchain
  • languages:-python-frameworks-/-libraries:-streamlit
  • rapidapi-(tripadvisor-api)-databases:-postgresql-apis:-tripadvisor-api-(via-rapidapi)
  • speechrecognition-platforms-/-cloud-services:-aws-(amazon-nova
Share this project:

Updates