Planning travel accommodations is time-consuming and often frustrating:
- Users need to visit multiple hotel booking sites.
- They must manually input dates, locations, and preferences repeatedly.
- Comparing options across different platforms is inefficient and frustrating.
- Natural language queries (like "Find hotels in Goa for next weekend") aren't supported by most booking sites.
This project creates an AI-powered hotel search assistant that:
- Understands natural language - Interprets casual requests like "Find 3-star hotels in Mumbai from May 5 to May 7 for 2 adults"
- Parse the requirement (location, check-in, check-out dates etc.) to create a MMT search URL
- Visit the URL and extract top 5 hotels
- For each hotel, find pricing and more info
- Summarize and presents key hotel information (prices, ratings, amenities)
- ๐๐ฟ๐ฒ๐๐๐ for multi-agent orchestration
- ๐๐ฟ๐ผ๐๐๐ฒ๐ฟ๐ฏ๐ฎ๐๐ฒ'๐ headless browser tool : to simulate human browsing and gather hotels data.
- ๐๐ฒ๐บ๐ถ๐ป๐ถ-๐ฎ.๐ฌ-๐ณ๐น๐ฎ๐๐ต-๐น๐ถ๐๐ฒ as the underlying LLM Model
๐ ๏ธ ๐๐๐ ๐๐๐ฃ๐ผ๐-๐๐ค๐ฌ๐๐ง๐๐ ๐๐ค๐ก๐ช๐ฉ๐๐ค๐ฃ
The app is built using CrewAI, a Python framework for multi-agent collaboration.
It primarily uses the below 3 Agents-
- ๐ฃ๐ฟ๐ผ๐บ๐ฝ๐ ๐ฃ๐ฎ๐ฟ๐๐ฒ๐ฟ ๐๐ด๐ฒ๐ป๐: Extracts structured fields (location, dates, number of guests) from the user query.
- ๐๐ผ๐๐ฒ๐น ๐ฆ๐ฒ๐ฎ๐ฟ๐ฐ๐ต ๐๐ด๐ฒ๐ป๐: Constructs a valid search URL on MakeMyTrip and returns it for downstream use.
- ๐ฅ๐ฒ๐๐๐น๐ ๐ฃ๐ฟ๐ฒ๐๐ฒ๐ป๐๐ฒ๐ฟ ๐๐ด๐ฒ๐ป๐: Scrapes and summarizes the hotel listings from the page into a concise, human-readable format.